Block Reference

Blocks are the building units of a workflow. Each block has a kind, a set of configuration fields, and a downstream output that later blocks can reference with expressions like {{input.field}} or {{lastOutput}}. This reference groups blocks by category and calls out the durable and AI blocks in detail.

Triggers

Triggers start a run. A workflow can expose several entries; an incoming file starts at its receiving trigger.

File input and extraction

Form Trigger receives fields and files from a shared page. Field names submittedAt and formMode are reserved for generated form metadata and cannot be used for text or File controls. Set the field name, type File, required state, single/multiple selection and accepted extensions. Form access defaults to workspace members; choose Anyone with the link explicitly before publishing a public form. Start supports the same file fields for manual runs. Webhook accepts multipart or raw binary. Enable Raw binary body for JSON or text bytes to become a file. Temporary file listeners use the same rule and keep listening after an unsupported body; multipart files do not require that option.

Each submission produces one item with json and binary. Form files use documento for a single selection or documento_0 for a multiple selection, including a selection with one file. Webhook repeated adjuntos[] parts use adjuntos0, adjuntos1; a configured multipart prefix data uses data0, data1. Raw Webhook uses the configured property or data.

Extract from File reads CSV, XLSX or text. Configure dataPropertyName to match the binary key and fileFormat as csv, xlsx or text. CSV extraction accepts up to 16 MiB, 10,000 data rows and 200 columns. XLSX accepts up to 20 sheets, 16 MiB per relevant XML entry, 32 MiB expanded XML and 10,000 worksheet rows including headers across the workbook. Both bound fields to 1,048,576 characters and serialized row output to 16 MiB; XLSX shared strings are limited to 100,000. An extraction limit fails the node with a bounded error; split or reduce the source file. These processing limits are separate from the 100 MiB upload limit. Plain-text extraction also accepts up to 16 MiB. CSV/plain-text source text must fit within 16 MiB after JSON escaping, excluding its surrounding quotes, before it is stored in execution history. HTTP Request offers binary and multipart request bodies. In multipart mode, Binary property selects the attachment to send; leaving it blank sends every binary property. Unrelated attachments are excluded when a property is selected. Code can preserve the binary descriptors when returning items. See the file walkthrough.

Selected HTTP file bodies share a 10-file/100 MiB budget across stored references and legacy binary values. Oversized legacy values are rejected before base64 decoding. Wait and Approval retain attached files while downstream item JSON contains the resume payload or decision metadata.

AI

The Agent block reasons over input, documents, tables, and retrieved context. Utility AI blocks add single-purpose model operations. All AI blocks run through BYOK — they reference a provider, model, and credential such as {{OPENAI_API_KEY}}.

Secret scanning

With secret scanning enabled, complete PEM private-key envelopes are detected as secrets: PKCS8, RSA, EC, DSA, OpenSSH and encrypted private keys. Guarded Agent external tool arguments are refused when they contain these secrets; guarded tool and final outputs redact the complete private block. Public-key and certificate envelopes are not classified as private keys.

Detection recognizes the envelope format, including JSON-escaped line endings. It does not validate cryptographic contents or guarantee detection of truncated blocks or arbitrary encodings. The existing permission and confirmation requirements still apply.

Imported single-call text consumers

Imported Basic LLM Chain versions 1–1.9 and Text Classifier versions 1 and 1.1 use the aiAgent kind with dedicated text operations. They require one supported OpenAI chat model source and an explicitly selected saved account. Chain makes one inference for one input item; an optional Item List version 1 splits raw text into separate items. Classifier makes one inference per item under one aggregate budget and keeps only items matching its single configured category. These profiles do not add Agent planning, tools, memory, fallback models or automatic format repair. Classifier 1.1 requires explicit sequential batch size one. Chain retains the version's prompt fields and literal text-message history; Classifier supports literal system templates with the original appended formatting instructions. An ordinary Agent version 2 can use Item List through its terminal formatter, preserving the original serialized argument wrapper and one numeric-key JSON item rather than Chain fan-out. Unsupported settings require migration. See the exact import profiles and recovery instructions.

End user memory

When the deployment enables end users, block settings include End user memory: connect memory resources to the workflow, mark the one block whose output is the reply to the end user, choose an End user website for Agent, Claude Code, Codex, and Website routine blocks, and run the draft as a test end user (choosing the website and running as a test end user need an explicit workspace Admin or Owner). While end users are enabled, the catalog hides the Memory preset because memory for your product's customers comes from memory resources. With end users off, the Memory preset stays in the catalog as before. Existing Memory blocks keep working either way.

Gemini

Search for Gemini or Google AI in the block catalog. The Gemini block uses the shared Agent runtime with Google Gemini selected, gemini-3.8-flash as its initial model, and a {{GOOGLE_AI_API_KEY}} credential reference. Connect a Gemini API key in Settings → BYOK, then select that workspace credential in the node. You can edit the model and configure prompts, tools, and the token budget in the same inspector as other Agent blocks.

Connecting Gmail or Google Drive does not provide a Gemini API key. Gemini itself requires no additional Workspace OAuth scopes. Before passing Workspace data to Gemini, verify the API project's billing tier and data-use settings against Google's Limited Use requirements; a connected key alone does not establish compliance.

Utility AI blocks

These are single-operation blocks on the aiAgent kind. Each picks a provider/model that supports the capability; without a credential they fall back to a consistent simulated mode and block deploy.

Configuring Embeddings

Embeddings shows provider, model, credential, input and optional dimensions. Select a saved workspace account in Credential, or choose + Add account... to create one. An OpenAI account asks for API Key, optional Organization ID and Base URL. The same account can be used by chat, embeddings and HTTP Request. Agent prompts, tools, guardrails and end-user memory do not apply to this block.

Website routine

Website routine (browserRoutine), under Developer Tools, runs a versioned script registered in the browser worker. Its steps use the customer's saved login without a model deciding each click. It runs only for a product end user; a manual member run cannot borrow a customer's session.

Choose End user website, save, then have an explicit workspace Admin or Owner select Authorize end user logins for this block. This block requires control authorization, including for lookups that use a website form. Set these fields in Config JSON:

{
  "browserSiteKey": "centris",
  "routineId": "centris.property_documents.collect",
  "routineVersion": 1,
  "parameters": {
    "propertyReference": "{{input.centrisNumber}}",
    "documentNames": "{{input.documentNames}}",
    "format": "{{input.format}}",
    "language": "{{input.language}}",
    "units": "{{input.units}}"
  }
}

A whole-value expression preserves its type: documentNames and units above are arrays. The document routine requires a Centris number or listing address, exact document names, client or broker format, fr or en, and one unit choice: ["imperial"] or ["metric"]. Missing or ambiguous business inputs return needs_input questions. An address must resolve to one complete listing result whose displayed address matches the query after Unicode NFC, case, and whitespace normalization. City, region, partial, or fuzzy results that show a different address need a Centris number. Multiple results, including historical listings at the same address, require the user to choose the Centris number before documents are opened or downloaded. Unknown routines, versions, or extra parameters are refused; a contract in the catalog is not sufficient unless its adapter is deployed. Workflow secrets are not expanded into these parameters.

An unavailable account page stops the routine without expiring the saved login. Retry after the site recovers; the routine still requires identity verification before it can use the account. A genuine signed-out account or a changed identity continues to require reconnection.

Keep these blocks in an invoked routine workflow so ordinary conversations do not require a website login. The block returns structured results; document bytes use the private routine artifact endpoint. Dry runs and single-step simulation do not visit the site. An in_doubt result needs reconciliation before another attempt at an external change. Website compatibility and each provider flow still require separate real-site verification.

Edit Fields Set

Search Edit Fields (Set) in Data to add, change or remove fields without writing code. Manual Mapping lets you add fields with a name, type and value. Types are String, Number, Boolean, Array and Object; values accept expressions such as {{$json.customerName}}. Names also accept expressions, resolved separately for each incoming item before the resulting path is validated; for example, {{$json.targetField}} chooses the field to update. Array and Object values use JSON; expressions inside quoted values preserve quotes, backslashes and line breaks from the input. JSON mode defines the output object directly; unquoted expressions preserve their JSON type, for example {"count": {{$json.count}}}.

Enable Include Other Input Fields to retain incoming fields; leave it off to output only the fields you define. Removing a field from the editor removes it from that defined output. Support Dot Notation turns customer.name into a nested property and users.0.name into the first array item's name, preserving the other items. Array paths require indices from 0 through 9999; larger indices and named array properties fail the step. Turn dot notation off to use a literal dotted key. Invalid conversions fail the step unless Ignore Type Conversion Errors is enabled, in which case the original resolved value is kept. Every incoming item is edited separately.

Generic HTTP Request

Search HTTP Request in Developer Tools. Set any URL and method, then configure Query parameters JSON, Headers JSON and Body template. Expressions can use input and previous node output. For authentication, select Predefined Credential Type, choose the Credential Type and a workspace Account, or add an account from that selector. Choose None when the endpoint needs no saved account.

For an OpenAI embeddings call, select POST, set https://api.openai.com/v1/embeddings, choose the OpenAI credential type and account, and supply a JSON body with model and input. Add a Content-Type: application/json header. OpenAI API keys and the optional organization header come from the encrypted account. Live HTTP requires the destination's live HTTP transport to be enabled; a simulated run prepares the request without sending it.

Logic and control flow

Parallel

Add Parallel from the Logic category, then connect at least two destination blocks to it. Those branches run concurrently, which is useful when they perform independent I/O. Their results retain the order of the connected branches. Each branch has its own output context; a branch cannot read another branch's output. To combine results, connect the ends of two or more branches to the same downstream block. That common block waits for all branches and can reference their outputs.

Ordinary fan-out still processes destinations in order. This includes connections from Condition; you can put a Parallel block under either the true or false output when that branch should fan out concurrently. Parallel coordinates workflow steps; it does not promise simultaneous CPU execution of function code.

Workspace subscription CLI calls using the same Claude Code or Codex credential lifecycle remain serialized by the existing credential lease. Parallel does not bypass that OAuth refresh protection.

Set "errorMode": "collect-errors" in the block's Config JSON to finish all branches and collect failures; the default is "fail-fast". A healthy branch stopped by fail-fast has result status "canceled", while the failure summary identifies the branch that actually failed.

If a branch pauses at a durable Wait, Human Task, or pending Website routine, completed siblings stay checkpointed and are not repeated when the run resumes. Do not put a Jenkins Trigger build and wait for result operation inside Parallel; Jenkins build-and-wait is not supported there. With fail-fast behavior, a failed branch does no further steps and the run waits for operations already in flight to finish. It cannot undo side effects already sent. Collect-errors behavior runs every branch and reports errors, but a failed branch prevents the common downstream block from running. Fatal CLI model-policy failures stop further steps in every branch, even with collect-errors, and fail the run without an automatic queue retry.

Loop

When a Wait, Human Task, or pending Website routine pauses a loop body, resumption continues with that iteration. Completed iterations and completed blocks of the current iteration are preserved. A pending Website routine resumes its original invocation before the loop advances to the next item.

Wait

The Wait block pauses a run durably — the run is parked with a waiting status and resumes without a live process holding it open, so it can wait minutes or days.

If the same selected branch also leads to other blocks, those pending destinations continue after the wait resumes. A destination that finished before the pause is not repeated. Fan-out destinations run as workflow progress allows; their execution is not simultaneous and has no guaranteed order.

A duration/date wait resumes automatically when the timer elapses — driven by a delayed job and backstopped by a scheduler sweep, so it still fires without Redis. An event wait resumes when a caller POSTs to a single-use resume hook (/api/hooks/resume/{token}). The token is crypto-random and single-use: a second POST, an expired token, or a cancelled run returns 410; an unknown token returns 404. Event payloads are JSON-only (non-JSON content types return 415) and size-capped (413 when exceeded).

Human in the Loop (Human Task)

The approval block is Confluye's Human Task. It pauses the run durably — the run parks with a waiting status, releases its worker lease, and holds no live process, so a task can sit open for minutes or days. A legacy approve/reject approval maps to a two-outcome Human Task; new tasks add named outcomes, candidates, forms/drafts, and explicit escalation and timeout behavior.

Assignment is independent of lifecycle: a task can be claimed, reassigned, started, paused, drafted, reminded, and escalated while it stays open. Only a terminal outcome consumes the linked wait token exactly once. Reminders and escalations notify without resuming the run, and an expired task is never treated as approval — it follows the timeout branch.

Humans work tasks from the Tasks inbox or the run detail's task card; the session routes are POST /api/human-tasks/{id}/actions (claim, unassign, reassign, start, pause, save-draft, complete, cancel). The legacy POST /api/runs/{id}/approval endpoint remains as a compatibility adapter over the task service. Every read/write is reauthorized against current membership, so a revoked user is rejected even on a task issued earlier; a concurrent second completion returns 409/ 410. External agents can monitor tasks read-only (see the Human Tasks API) but can never complete a human approval.

For detection, diagnosis, and safe recovery of parked, orphaned, or racing tasks, see the Human Task recovery runbook.

Data

Imported Limit, Sort, Aggregate, Split Out, Remove Duplicates, Rename Keys and Summarize use their original parameter profiles instead of generic array approximations. Inspect the retained parameters in Config JSON. Limit and Sort preserve ordered native items and binary metadata; Aggregate returns one item, Split Out expands fields, and Summarize returns summaries. Empty collections stop downstream activation, including when the imported data node is disabled. Expressions and unsupported modes show an explicit migration blocker. Remove Duplicates does not persist history; Sort code mode is unavailable; Rename Keys regex processing uses a bounded subset without alternation, requires ^ before variable-length repetition and shares a total work budget across items. See versions, options and repair.

En español: los perfiles importados ejecutan sus parámetros originales y rechazan las opciones que no admiten. Una colección vacía detiene los nodos siguientes. La deduplicación no conserva historial y las expresiones regulares tienen límites. Consulta las versiones y reparaciones en la guía de importación.

For an image column, Table Write accepts a minimal Workspace File reference or a binary descriptor from an authenticated workflow input. Preserve the object's type by leaving the expression unquoted:

{"Name":"Product","Photo": {{input.items[0].binary.Photo}}}

To try this with a manual workflow:

1. Create a table with Name (text) and Photo (image).

2. Add a file input named documento to Start. Select an image in Resources and wait for its ready status.

3. Add Table Write, select your table, and enter these Values:

   {
     "Name": "Product A",
     "Photo": {{input.items[0].binary.documento}}
   }
   

4. Run the workflow and open Tables to inspect the saved record and image.

5. In Table Read, select that table. Each output item exposes the record in json and the image in binary.Photo.

Use the actual binary property name from your producer (for example documento from a single-file form). The server verifies the file and reduces the cell to {kind, fileId}. Table Read keeps its legacy rows output and also emits items with each row in json and its images in binary, keyed by column name. An HTTP Request with binary property Photo can send those bytes in this run or a later run. The authenticated principal needs execution and restricted data access. The existing ten-file / 100 MiB binary budget applies; oversized reads fail explicitly. Use safe binary field names for image columns.

Response

The Response block sets the status, headers, and body of the reply to a webhook (or chat) caller — then the run continues (respond-and-continue). The first Response wins; any later Response is a no-op with a warning in the run log.

For same-origin safety, the content type is restricted to a safe set (JSON, plain text, CSV, markdown, images, PDF, and similar). Executable types like HTML are only served as a non-executable download (Content-Disposition: attachment), and X-Content-Type-Options: nosniff is always applied. The status must be in the 200–599 range. If the run hasn't produced a Response within a ~30-second budget, the webhook returns 202 with a processing status and keeps running; if the run enters a waiting state first, it returns 202 with status: "waiting" and the runId. On a manual or scheduled run (no HTTP caller), the Response body simply becomes the block's output.

Knowledge

knowledgeSearch and knowledgeDocument are separate nodes: search only reads, and the document node performs content lifecycle through the canonical document capabilities. A knowledgeDocument create/update can optionally park on a generation-aware searchable wait and resume with the terminal indexing state (indexed, keyword_only, or failed) as its output; a stale generation cannot wake a newer mutation. "Delete" means reversible move-to-trash with immediate exclusion from retrieval — permanent purge stays an administrator-only API operation. See the Knowledge Base guide.

Integrations

Integration blocks call third-party APIs. Confluye ships a broad catalog (GitHub, Slack, Gmail, Google Sheets/Slides/Drive/Calendar, Jira, Confluence, HubSpot, Salesforce, Notion, Linear, and many more), grouped in the canvas under Developer Tools, Communication, and Integrations.

Connectors are increasingly backed by integration manifests: each provider declares typed operations (input/output schemas), credential requirements, and dynamic option loaders, so the block exposes a real operation selector instead of a raw HTTP facade. Each provider/operation carries a status — unavailable, preview, or stable — and deploy is blocked for unavailable operations. An operator kill-switch env var can additionally disable a provider. Credential material is redacted from run logs.

Notion adds 15 native page/database/block/user operations, Supabase adds seven row/function operations, and Pinecone adds five index-management operations. Each uses a selected saved account. Supabase update/delete require column filters; Pinecone vector operations are not included. See account setup and operation limits.

Airtable adds eight base/schema/record operations, HubSpot adds six CRM operations across contacts/companies/deals/tickets, and Asana and ClickUp each add eight project/task operations. They use saved personal/private-app tokens. Lists return one page and expose the provider's cursor or page number. See token setup, pagination and supported scope.

GitLab adds 23 project/issue/comment/file/release operations, Discord adds 16 bot/channel/message/member operations, SendGrid adds 12 contact/list/email operations, and Brevo adds 18 contact/list/attribute/sender/email operations. GitLab text writes create commits; Discord uses a bot token and preserves string IDs. SendGrid contact writes queue jobs, and its sample contact read is limited to 50. Brevo supports existing transactional templates and explicit contact upserts. See credentials, permissions and operation limits.

Todoist adds 39 operations across tasks, projects, sections, comments, labels and reminders using the unified API v1. Lists return one page with next_cursor; IDs remain strings. Movement and comments use an explicit target kind; reminders have separate relative and absolute timing operations. See Todoist account setup and limits.

Linear adds eleven operations for issue management, comments, link attachments and team/user/state discovery. It uses GraphQL variables, preserves data and cursor page information, and rejects HTTP 200 responses with GraphQL errors or unconfirmed mutations. See Linear credentials, field choices and limits.

Trello native API blocks provide 40 operations for boards, lists, cards, comments, checklists/items, labels and card members. Select a saved trelloApi account with both API key and token. Create boards in an explicit workspace; archive with Closed instead of permanent board/card deletion. Partial updates require a selected field and preserve false values. See Integrations and credentials for paging, account setup and unsupported fields.

Freshdesk adds 28 ticket/conversation/contact/company/agent/group operations. Save a freshdeskApi account with its tenant subdomain and API key. IDs are numeric; notes default to private. Tickets and contacts have soft-delete and restore operations. See Freshdesk setup and limits and typed parameter expressions.

Zendesk adds 35 ticket/comment/tag, suspended-ticket, user, organization, group and ticket-field operations. Select a saved zendeskOAuth2Api account with subdomain and a valid OAuth token. Comments default to private notes, and suspended-ticket recovery changes its requester to the authenticated agent. See Zendesk setup and limits.

Stripe adds 31 operations for balance, customers, charge metadata, coupons, products/prices, payment methods, meter events and payment-intent status reads. Select a saved stripeApi key for the intended account. Amounts use integer smallest currency units; zero-price and false updates are preserved. Payment collection and card tokenization are outside this adapter. See Stripe setup and limits.

Mailchimp adds 23 account/audience/interest-group, member/tag and existing-campaign operations. Select a saved API token and identify existing members by their returned ID or lowercase-email MD5 hash. Creation requires an explicit subscription status. Campaign Send uses the configured recipients. See Mailchimp setup and limits.

Twilio adds 18 message/call/recording/number/balance operations. Select a saved twilioApi account with its Account SID plus Auth Token or API key pair. Sending uses a sender and body/media or Content template; calls need TwiML or an instruction URL. Writes may incur charges. See Twilio setup and limits.

Pipedrive adds 36 identity and CRM operations using API v2 for deals, persons, organizations, activities, products, pipelines and stages. Select a saved pipedriveApi token; updates require a changed field and preserve false/zero. Lists use a cursor and keep the success/data envelope. See Pipedrive setup and limits.

WordPress adds 29 core REST operations for posts, pages, categories, tags, comments and profile reads/updates. Save an HTTPS site URL, username and application password as wordpressApi. Creation defaults posts/pages to Draft; deletion distinguishes Trash from permanent removal. Lists use page/per-page inputs. See WordPress setup and limits.

WooCommerce adds 30 REST v3 operations for products, orders, customers, coupons and product categories/tags. Select a saved wooCommerceApi store URL and consumer key/secret pair. Products default to Publish; orders start Pending/unpaid. Prices are decimal strings, and deleting customers/terms is permanent. See WooCommerce setup and limits.

Baserow adds 11 operations for discovering tables/fields, reading and changing rows, moving rows and batch writes. Select a baserowApi HTTPS host/database-token account, use numeric table/row IDs and exact column names in Fields JSON. Lists return one page; batches accept up to 200 items. See Baserow setup and limits.

Contentful adds ten delivery/preview reads for the selected space, content types, entries, assets, locales and tags. Choose a contentfulApi account with the matching token/source and US/EU region. Environment defaults to master; collection responses retain items, linked includes and pagination metadata. See Contentful setup and limits.

MailerLite adds 23 current-API operations for subscribers, groups, custom fields, import progress and subscriber ecommerce reports. Select a mailerLiteApi token, keep IDs as text and distinguish additive upsert groups from replacement groups on update. Import and scheduled forgetting are asynchronous. See MailerLite setup and limits.

Fixed-field collections use Add item. Untouched optional fields stay omitted; after editing one, use Omit to remove it explicitly. Clearing text preserves an explicitly empty value when the operation permits it. Required fields still need valid values; numbers are committed when you leave their row field, and false/zero defaults keep their types. Save the node after editing.

For fields that accept several options, such as Gmail Label Ids, select an account and open Choose label ids. Search by name or ID, then check the options you need. Selected labels remain visible when the list is collapsed; use the remove button on a selected label to deselect it. Searching does not clear earlier selections. Save the node to keep your choices. If options fail to load, use Retry.

Real integration operations

These operations are declared in the integration manifests and are the provider-specific operations the block selector exposes.

Write operations carry side-effect metadata and may require confirmation. Some read/list/search operations are still marked preview in the manifest when they are backed by declarative HTTP rather than a dedicated runtime executor.

1Password (authenticated request)

The 1Password block (onePasswordRequest) sends one HTTPS request that is authenticated by a 1Password field a workspace Owner granted to the workflow for an end user. The workflow stores only the credential's reference and the one host it may reach. Confluye adds the credential at the last moment, after it checks the grant again, and the block's output is the response with the credential removed.

Use it when a workflow must call an API for one of your product's end users with that end user's key. An end user is whoever owns the assistant in your product, such as a real-estate agent in a CRM, not the people that end user serves. Find it by searching 1Password in the block catalog, where it is listed under Developer Tools. It runs only for an end user: a run your product starts for that end user. A manual test, a schedule or any run without an end user is refused with a message saying so; there is no fallback to a shared vault. A dry run or single-step simulation shows the request it would build and reads nothing: it neither contacts 1Password nor sends the request. The block cannot browse or search a vault and never outputs the credential. Confluye has checked this block against a test fixture only, not against a live 1Password vault; check it with a synthetic vault and item before you rely on it.

Before you add the block:

1. A workspace Owner connects a vault and grants one item field to this workflow and end user in Settings → Environments → Private credentials (1Password). See Credentials from a private 1Password vault.
2. The 1Password item lists the API's host as its website.

The block shows what is missing before you run it: no credential chosen, no allowed host, a URL on another host, an http:// URL, a header that Confluye controls (such as Host), or a body that cannot carry the credential. Deploy is blocked until those are fixed.

Output. response.status, response.headers (content type, location and a few others) and response.body (parsed when the server answers JSON), plus request with the credential shown as {{REDACTED}}. A server that echoes the credential back has it replaced by [REDACTED SECRET]. A status of 400 or more fails the block, like an HTTP Request block. Redirects are not followed: a 3xx answer is returned as it is, with its location header.

Example. Look up a client record in a CRM and branch on the answer:

1. Manual trigger receives { "customerId": "c-77" } from your product.
2. 1Password with method GET, URL https://api.crm.example/v1/customers/{{input.customerId}}, credential op:…/…/password, allowed host api.crm.example and bearer placement.
3. Condition checks response.body.name; Output returns the result.

When a request is refused. The block fails and nothing is sent. The messages say only what a person can act on:

A grant is checked when the run starts reading the credential and again immediately before the request is sent, so revoking a grant or disabling the vault stops a request that has not gone out; a request already sent is not recalled.

A write that may have gone out is never repeated. With Retry on fail on, a GET or HEAD that fails is tried again, and so is any request that was refused before it left Confluye (a grant check, the credential provider or the network safety check). A POST, PUT, PATCH or DELETE that was already sent and then failed, because the response was lost, could not be read or had a status of 400 or more, is not retried: Confluye cannot tell whether the other system applied it, and sending it again could apply it twice. The block fails once with the message The outcome of this … request is unknown. Check the other system for the change before you run the block again, and use an error output to handle that failure.

API Batch

A apiBatch block fans out one HTTP request per item in an input list, with limited concurrency, then aggregates the results.

Results are aggregated as { results[], errors[], stats } in input order. All request URLs pass through the SSRF egress guard (see below).

A2A (agent-to-agent)

The a2aCall block invokes an external agent over the A2A protocol and uses its response downstream.

Confluye also publishes an agent card and an invoke endpoint for your own deployed workflows, so another Confluye (or any A2A client) can call them. The agent card lists deployed workflows as skills; invocation authenticates with a workspace API key and is opt-in per workspace. Sync invocation maps to a run plus its Response and is limited to workflows without waits. Async invocation returns 202 with a task id and supports workflows that pause on Wait or approval blocks.

See A2A for async tasks, polling, callbacks, signatures, and task expiry.

Call workflow (workflowCall)

Runs another Confluye workflow in the same workspace. Config: workflowId (target id or imported n8n id), optional workflowName, input (default {{lastOutput}} on n8n import). Deploy stays blocked while the target is still a placeholder or missing. Runtime resolves imported n8n ids after the child bundle is upserted.

Imported n8n Code nodes are function nodes with n8nType ending in code. When FLUXUS_ENABLE_WORKFLOW_CODE=true, Code and general JavaScript expressions run in a fresh, one-use Linux child with the installed seccomp policy, bounded serialized data, and no application credentials, filesystem, network, or subprocess access. The legacy FLUXUS_ALLOW_UNSAFE_WORKFLOW_CODE variable does not select a backend and is ignored. If the operator flag is off, or Linux and the policy are unavailable, deployment and execution stay blocked with a recovery diagnostic; there is no in-process fallback. See Importing n8n.

Imported Agent models, Code Tools and debugging controls

Supported imported model subnodes attach through ai_languageModel; they do not run as ordinary workflow steps. Ollama version 1 requires an explicitly selected saved account and matching reviewed HTTPS n8nOllamaOrigin on the Agent. Its native /api/chat adapter makes one inference attempt; this does not inherit the generic local-provider loopback settings below.

The bounded Model Selector version 1 connects two indexed OpenAI Chat 1.3 sources to one V2 Agent. Review both routes and one shared saved target account before activation. Strict string rules select one route for that Agent activation; no match stops and provider failure does not select another model. Other profiles and waiting tool confirmation require migration; see Importing n8n.

Imported Code Tool versions 1 through 1.3 attach through ai_tool and run only when selected. Their synchronous JavaScript uses the existing isolated Linux opt-in and accepts only an optional literal query; the model cannot supply code. This profile differs from an ordinary main-flow Code block. A validated DebugHelper version 1 doNothing block emits zero items and stops its downstream path; its enabled profile requires one linear flow. See the exact versions, account setup, unsupported options and recovery instructions in Importing n8n.

Egress protection (SSRF)

Any user-supplied URL — the a2aCall endpoint, apiBatch targets, and local provider base URLs — passes through a single egress guard that blocks private, link-local, and loopback ranges and *.railway.internal (except an explicit allow-list), and re-validates redirects against the same rules. Local providers (Ollama/vLLM) are the one case where loopback is intentionally allowed, via their host allow-list.

Next steps

Intercom API blocks provide 38 contact, company, note, tag, segment and subscription operations. Use text IDs, the selected regional account and explicit consent for subscription changes. Read Intercom account setup and node limits before configuring writes.

ActiveCampaign API blocks provide 38 contact, list, tag, CRM account, association and custom-field operations. Configure the full Developer-tab API URL and key, distinguish resource IDs from association IDs, and select subscription status explicitly. See ActiveCampaign setup, fields and side effects.

NocoDB API v3 blocks expose 24 record, relationship, attachment, button and metadata-read operations. Use text primary keys and inspect computed-field warnings; bulk writes accept up to ten records. See NocoDB setup, API versions and node limits.

Kit (ConvertKit) API v4 blocks expose 40 account, custom-field, form, sequence, subscriber and tag operations. Use a v4 personal API key, existing subscriber IDs and explicit sequence settings; inspect warnings and asynchronous field updates. See Kit setup and node behavior.

Bitly API v4 blocks expose 33 link, custom-alias, group, domain-read and analytics operations. Use a saved access token and text IDs such as bit.ly/abc123; explicit false archive flags and empty tag arrays are retained. See Bitly setup, pagination and write behavior.

Strapi 5 blocks expose 14 document, single-type, publication and media operations. Use a Content API token and text documentId values; native writes explicitly default to draft where Draft & Publish is enabled. See Strapi setup, version limits and publication behavior.

Clockify API v1 blocks expose 38 workspace, client, project, task, tag, user and time-entry operations. Choose the saved key's region, text resource IDs, explicit project visibility and UTC timestamps; lists return one page. See Clockify setup, timer behavior and limits.

Webflow adds 33 Data API v2 operations, separating staged CMS changes from live updates, unpublishing and publishing. Use the selected token and text IDs; inspect publishing errors. See Webflow setup and limits.

Help Scout adds 26 Mailbox API v2 operations with an existing OAuth access token. Conversation and reply writes may send email; token renewal is manual. See Help Scout setup and limits.

Mautic adds 28 REST operations with an HTTPS origin and API-enabled Basic Auth account. Memberships, points and sends can cause automation or outbound mail. See Mautic setup and limits.

Coda adds 27 current API operations for docs, page metadata, rows and other resources. Writes may return an asynchronous request ID and button pushes may run automation. See Coda setup and limits.

Monday.com adds 47 operations with typed GraphQL variables and API version 2026-07. Column values use column IDs; writes can change boards or notify subscribers. See Monday.com setup and limits.

Calendly adds 40 REST operations, including availability, scheduled events and booking/cancellation. Native actions do not implement the reference n8n trigger. See Calendly setup and limits.

Typeform adds 24 REST operations. Whole-form replacement can delete omitted fields and response deletion is queued; webhook definitions do not implement a trigger. See Typeform setup and limits.

Shopify, Salesforce and Zoho CRM

Use these native service blocks with a selected saved account. Shopify setup and limits covers the pinned GraphQL API, shop-bound token, typed GIDs, explicit inventory idempotency keys and asynchronous collection changes. Salesforce setup and limits covers My Domain instance binding, six standard object types, structured queries, text IDs and version-specific cursors. Zoho CRM setup and limits covers region-bound API v8, scopes, text IDs, bounded batches and automation effects. These operations do not add the n8n Shopify or Salesforce triggers or automatic credential/token renewal.

Batch record inputs that accept custom field names use a JSON editor, including nested lookup IDs and empty/false/zero values. Fixed field lists keep their row controls. Enter a full JSON array for a batch, or map a whole expression in that field; save before running. Required record IDs remain visible to validation even when the batch supports other custom keys.

Dropbox, Box and Zoom

Use a selected saved account with these native service blocks. Dropbox setup and limits covers JSON RPC, stored namespace/member routing, path IDs and pending share jobs. Box setup and limits covers decimal IDs, file/folder selection, collaboration/shared-link effects and ordered metadata patches. Zoom setup and limits covers OAuth scopes, scheduling/registrant effects and recording metadata without host start URLs or download tokens. These actions do not add automatic pagination, token renewal, binary transfers or webhook triggers. Connection checks read identity and do not certify writes.

Discourse, Copper and Netlify

Discourse setup and limits explains the fixed acting user, authenticated connection check, post modes, moderation permissions and queued publication. Copper setup and limits explains key-owner email, text IDs, resource-specific CRM fields and bounded searches. Netlify setup and limits explains site/build/deploy effects, metadata-only responses, form/submission arrays and read-only DNS. Select a saved account, review the operation's effects and inspect provider state before repeating an uncertain write. These actions do not add automatic pagination, OAuth renewal or workflow triggers; connection checks do not certify writes.

DeepL, Cloudflare and BambooHR

DeepL setup and limits covers explicit Free/Pro authority, ordered text arrays, empty Context, typed options and billable translation. Cloudflare setup and limits covers saved zone/account verification, finite A/AAAA/CNAME DNS actions, explicit PATCH tuples, destructive preflights and uncertain deletion acknowledgement; shipped n8n certificate actions remain unsupported. BambooHR setup and limits covers the saved company, self-ID limits, confidential field projections and the published directory's incomplete roster coverage. Select a saved workspace account. Checks prove only their performed reads. These profiles add no OAuth renewal, triggers or automatic n8n conversion. Local HTTP is simulated; live effects, hosted Review, organization and durable behavior remain untested.

Mailgun, Microsoft Excel and OpenWeatherMap

Mailgun setup and limits covers the saved region/domain, scoped read check, queued mail acceptance, templates and finite logs. Microsoft Excel setup and limits covers the saved business workbook, delegated permissions, ordered row-schema checks, positional row effects, typed matrices and persistent sessionless changes. OpenWeatherMap setup and limits covers coordinates, units, timestamp/precipitation semantics, ambiguous geocoding and incomplete forecasts. Select a saved workspace account; checks do not prove all permissions. These profiles add no token renewal, triggers or automatic n8n conversion. HTTP is simulated locally; live effects, Review, organization and durable behavior remain untested.

Pushover, Agile CRM and Grafana

Pushover setup and limits explains application and recipient keys, group/device fanout, quiet hours and queued acceptance. Agile CRM setup and limits covers saved Basic credentials, exact IDs, contact/company kind checks and automation effects. Grafana setup and limits covers the legacy API profile, scoped team-read check, access changes and partial metadata. Select the saved workspace account and inspect permissions and write effects. These actions do not add triggers, OAuth renewal or automatic n8n conversion. Provider HTTP is simulated; live effects and hosted Review remain unverified.

PagerDuty, Sendy and Raindrop

PagerDuty setup and limits explains the saved REST token/From user and explicit incident, alert and responder effects. Sendy setup and limits covers the saved installation URL, form authentication, inferred brand collection shape, subscriber mail effects and draft-only campaigns. Raindrop setup and limits covers existing tokens, collection scopes, page fetching, sharing and distinct merge tags. Select a saved workspace account and review writes before running. A successful check does not certify all permissions or effects; these actions do not add token renewal, triggers or automatic n8n conversion.

Pushbullet, UptimeRobot and Sentry

Pushbullet setup and limits covers the saved token, explicit note/link recipient, omitted empty fields, deleted stubs and notification effects. UptimeRobot setup and limits covers current v3 Bearer tokens, HTTP monitors, once-only maintenance, alert assignments and explicit public status-page choices. Sentry setup and limits covers scoped organization reads, current organization issue paths, cursor pages and bounded issue-state changes. Use saved workspace accounts and review writes before running. Read checks do not certify write authority; these actions do not add automatic pagination, token renewal or triggers.

GetResponse, Freshservice and Lemlist

GetResponse setup and limits covers the saved SMB key, queued contact acceptance, replacement arrays and custom-field choices. Freshservice setup and limits covers the saved tenant/key, ticket requester selection, explicit note privacy and requester visibility. Lemlist setup and limits covers the saved team key and required acknowledgment that adding leads may auto-launch outreach. Select the saved account and review effects before running. These actions do not implement the reference's GetResponse/Lemlist triggers. Lists read one bounded page, connection checks do not certify writes, and uncertain writes are not retried automatically.

Grist, Harvest and Microsoft To Do

Grist setup and limits covers the saved HTTPS host/key, anonymous-profile refusal, dynamic cell maps, safe record IDs, upsert matching and bounded reads. Harvest setup and limits covers the saved account/application headers, resource-specific fields, timer modes, decimal costs and locked-expense checks. Microsoft To Do setup and limits covers delegated Tasks scopes, opaque IDs, local timezones, recurrence/reminders and resource-bound continuation links. Select a saved account and review writes before running. Connection checks do not certify writes. These actions do not add automatic pagination, token renewal or workflow triggers.

Mailjet, Demio and Customer.io

Mailjet setup and limits covers the saved Email key pair/region/sandbox mode, mode-matching send acknowledgement, template/contact reads and changes, and acceptance without delivery proof. Demio setup and limits covers the saved key pair, explicit event/session IDs, quota-consuming reads, registration notifications and bounded participant reports. Customer.io setup and limits distinguishes Track and App credentials, provider environment versus local workspace, typed profile/event data, audience effects and finite metrics. Select the matching saved workspace profile and review effects. These actions add no trigger lifecycle, token renewal or automatic n8n parameter/credential conversion. Provider HTTP is simulated locally; live effects, hosted Review, organization and durable behavior remain untested.

Mattermost, Jotform and Segment

Mattermost setup and limits covers twelve actions with a saved public installation, exact acting user and active-user preflight before every action. Post/reaction writes require explicit acknowledgement and can notify people. Jotform setup and limits covers seven regional reads, exact decimal text IDs, bounded pages and explicit sensitive-answer inclusion; it adds no Jotform Trigger. Segment setup and limits covers six Tracking actions with explicit event IDs, destination choice and effects acknowledgement. Segment accounts stay saved/unverified with no connection event; HTTP 200 proves only a transport receipt. Whole-expression IDs and exact Segment decimal fields retain their original types, so numeric values fail strict validation rather than being silently converted to text. No full-node/import/trigger parity is claimed. Provider HTTP is simulated locally; live authentication/effects, hosted Review, new-node organization and durable behavior remain untested.

NASA NeoWs, Metabase and Google Tasks

NASA setup and limits covers three quota-consuming NeoWs reads with exact SPK text IDs, an explicit seven-day feed range and one returned page. Historical fixture projection does not certify current API compatibility or scientific suitability. Metabase setup and limits covers thirteen metadata reads and one saved-question execution using a selected group API key on a public installation. Query execution needs an explicit acknowledgement; selected rows/columns limit output, not compute or datasource writes. Google Tasks setup and limits covers twelve task/list actions with a manually saved token and explicit local read/write declaration. Saving does not contact Google; Test connection explicitly reads at most one list. Writes require effects acknowledgement; deletion can remove assigned originals in Docs/Chat, and clear hides every completed task. Exact IDs retain their types through expressions. No automatic pagination, retries, token renewal or full reference-node/import parity is claimed. Local provider responses are simulated; live behavior, hosted Review, new-node organization and durable behavior remain untested.

Hacker News and CircleCI

Hacker News setup and limits covers public official API reads without a saved account, exact text IDs, inert HTML and a local slice of mutable feed IDs. Map a feed ID into a later item node explicitly; no hydration or Algolia search is added. CircleCI setup and limits covers ten Cloud API v2 actions using a saved personal provider token. Saving does not contact CircleCI; Test connection is one explicit identity read. Pipeline triggers and cancellation require acknowledgement plus workflow confirmation. Creation/cancellation acceptance does not prove execution completed, and an uncertain write is not retried automatically. Local responses are simulated; live provider, hosted Review, organization and durable behavior remain unverified.

Brandfetch

Brandfetch setup and limits covers six cached-only Brand API projections using an explicitly selected saved key. Choose domain or Brand ID; missing, null and empty values remain distinct. Saving makes no request; Test connection explicitly reads Viewer once without proving brand access or credits. Cached hits can consume credits, and an unindexed result does not start a crawl. Asset URLs are inert text and are never downloaded. Search, automatic collection and automatic n8n conversion are excluded. Local responses are simulated; live provider, hosted Review, organization and durable behavior remain unverified.

Action Network and Affinity CRM

Action Network setup and limits covers 26 reads and thirteen writes using a saved list-scoped key and explicit list-type declaration. Saving stays unverified without a connection test. Writes require effect acknowledgements plus separate workflow confirmation and may change subscriptions, replace action records or affect automation; imported events/petitions are record-only. Affinity setup and limits covers eleven reads and two scalar list-field writes using a v2 Bearer key and exact expected tenant. Saving makes no request; Test connection separately checks whoami once. Scalar writes compare an explicit typed prior value after confirmation; GET/POST is not atomic, and empty text, zero and explicit clearing remain distinct. Both profiles use finite single-attempt phases even with Retry On Fail enabled. Investigate uncertain writes before rerunning. No automatic n8n conversion or live trigger is added. Provider responses are simulated locally; live behavior, hosted Review, organization and durable persistence remain unverified.

Adalo

Adalo setup and limits covers four Collections API actions: get, list one page, create and delete. Save the key and fixed app ID without a connection test, then select an explicit collection and declared scalar field profile in each node. Typed slot assignments preserve empty text, zero and false. Reads require personal-data acknowledgement; writes also require effect acknowledgements and separate workflow confirmation. Collection type, defaults, downstream effects and field persistence remain unverified. Explicit failure markers in a deletion response stop the action even with HTTP success. Continuations are scope-bound candidates without automatic pagination, and uncertain writes are never retried automatically. No update, bulk action, relationship/authentication support or n8n import conversion is added. Local provider responses are simulated; live provider, Review, organization and durable persistence remain untested.

APITemplate.io, Airtop and Autopilot Journeys

APITemplate.io provides five v2 quota/list/generation operations. Use typed PDF slots in the JSON editor or fixed image text rows; generation requires quota/assets/storage acknowledgements and workflow confirmation. Airtop reads an explicit session, lists its window IDs or closes one window. Every action needs sensitive-metadata acknowledgement and workflow confirmation; closing additionally acknowledges browser control and does not stop session billing. Autopilot Journeys provides eight legacy contact/list/journey actions, separate from Ortto CDP. Reads acknowledge data access; writes also acknowledge automation effects and require confirmation. Optional names/titles need explicit inclusion and acknowledgement.

Save each native API key without a provider request or connection test and explicitly select the saved account in the block. Required acknowledgements start unchecked. These profiles use one attempt even with Retry On Fail enabled. Cancellation after dispatch does not undo a write; investigate possible effects before manually rerunning. Provider execution deadlines exclude credential and confirmation preparation. These actions do not add n8n import conversion or live triggers. Real provider behavior, hosted Review, organization scope and durable persistence remain untested.

Acuity Scheduling and Bannerbear

Acuity setup and limits covers seven finite account/calendar/availability/appointment actions using a saved Basic User ID and key. All actions require personal-data acknowledgement and workflow confirmation. Cancellation and rescheduling additionally require explicit effects, notifications, costs and noEmail choices; availability is a snapshot, and cancellation cannot be undone through this endpoint. Bannerbear setup and limits covers four V5 image/template actions with a distinct V5 key. Every action acknowledges data access and requires confirmation; generation also acknowledges quota, possible template asset retrieval and public storage. Typed modifications preserve empty text, false and zero; 202 acceptance can leave an image pending.

Save account makes no provider request and Test connection is unavailable for both profiles. Select the exact saved native account. Both use finite outputs, one attempt even with Retry On Fail, execution deadlines after credential/confirmation preparation, and explicit later actions for additional pages or status. Cancellation after dispatch can leave a write applied. Neither profile adds automatic n8n conversion or live triggers. Real provider, hosted Review, organization and durable behavior remain untested.

Beeminder

Beeminder setup and limits covers goal metadata, one datapoint page and datapoint creation with a selected saved personal auth token. Save makes no request and Test connection is unavailable. Every action needs personal-data acknowledgement and separate workflow confirmation; creation additionally acknowledges commitment and financial consequences. Value and optional Unix timestamp are typed numbers; explicit empty comment, zero, negative values and omitted fields retain their meanings. Replies are bounded to 25 datapoints, even when per=0 requests all. Each invocation makes one attempt despite Retry On Fail. A cancelled or uncertain creation may have applied; inspect the goal before another invocation. No goal/pledge administration, OAuth or automatic n8n conversion is added. Real provider, Review, organization and durable behavior remain untested.

Chargebee and Bubble

Chargebee setup and limits covers eight finite API v2 customer, invoice and subscription actions with a saved exact site/key. All actions acknowledge personal data and require separate confirmation. Cancellation checks current subscription/contract state before one POST and requires explicit scheduling and financial choices; a concurrent change can still occur. PDF links are sensitive capabilities, never downloaded. Deletion schedules a permanent cascade, requires external prerequisite review and does not delete associated usage events or prove completion.

Bubble setup and limits covers five Data API actions using a saved app, explicit live/development environment and declared user/admin token class. Match the saved app/environment in each block and declare exact scalar fields and read selection. Writes use strict rawFieldsJSON; false, zero and empty text remain values, declared nullable reads preserve null and null writes are unsupported. All actions require data acknowledgement and confirmation; admin additionally acknowledges privacy bypass, writes effects and deletion destructive consequences.

Save sends no request and Test connection is unavailable for both profiles. Each invocation makes one attempt despite Retry On Fail; pagination is explicit and returned URLs are inert. A lost or cancelled write may have applied. Inspect the resource before rerunning. No automatic n8n conversion, live trigger or rollback is added. Real provider, hosted Review, organization scope and durable behavior remain untested.

Currents.dev

Currents.dev setup and limits covers three test-analytics reads with a saved organization API key: a project page, one project, or a run page. Save sends no request and Test connection is unavailable. Every action needs personal-data acknowledgement and separate confirmation. Lists use typed limit 1–25 and one cursor direction; an edited empty cursor differs from omission. Run search preserves surrounding spaces. Optional run fields can remain absent, and no ID or cursor is invented. Each invocation makes one attempt despite Retry On Fail; no automatic pagination, trigger or n8n conversion is added. Provider, Review, organization and durable behavior remain untested.

CoinGecko

CoinGecko setup and limits covers three Demo API reads: supported quote currencies, explicit-ID USD prices and one USD market page. Save sends no request and Test connection is unavailable. Select the saved Demo account and confirm each invocation. IDs require a strict JSON array or a complete typed array expression; markets require explicit numeric page and perPage. Missing prices remain absent, nullable changes retain null and zero stays numeric. Each invocation makes one attempt despite Retry On Fail. No Pro or anonymous fallback, automatic pagination, trading, trigger or n8n conversion is added; price freshness and live provider behavior remain unverified.

Disqus

Disqus setup and limits covers four server-secret reads: a forum, threads, posts or categories. Save sends no request and Test connection is unavailable. Select the saved native account, enter the exact forum shortname, acknowledge personal data and confirm each invocation separately. Lists use a typed limit, optional order and an explicit cursor for each page. Structured author/email/IP/media fields are omitted, but selected raw_message remains untrusted text and may contain personal data or HTML. Null, false, zero, empty and missing values retain their declared meanings. Each invocation makes one attempt despite Retry On Fail. No automatic pagination, write, moderation, OAuth or n8n conversion is added. Real provider, Review, organization and durable behavior remain untested.

DHL

DHL setup and limits covers one production shipment-tracking read. Select a saved dhlApi subscription key, enter the exact tracking number, review any optional postal enrichment, acknowledge personal data and confirm separately. Save sends no request and Test connection is unavailable. Offset and limit are typed numbers with local bounds; each invocation makes one attempt without automatic pagination. Returned dates and service/status codes remain bounded raw text. Detailed addresses, locations and proof of delivery are omitted, while remaining tracking metadata may still be personal. Real provider, Review, organization and durable behavior remain untested.

Cockpit

Cockpit setup and limits covers two native Cockpit 2.14.1 content reads. Select a saved cockpitContentApi account, the model and 1–8 exact scalar field names. Lists also require an explicit numeric limit of 1–25 and read one first page only. Missing selected fields remain absent, while null, false and zero retain their types. Item reads can return literal false. The personal-data acknowledgement starts unchecked and each invocation needs separate confirmation. Save sends no request and Test connection is unavailable. Each invocation makes one attempt despite Retry On Fail. Nested selected content, arbitrary filters, writes, population, legacy query-token routes and automatic pagination are unsupported. Real provider, Review, organization and durable behavior remain untested.

Drift

Drift setup and limits covers two native reads with driftApi. Contact reads require an exact decimal string ID and an explicit JSON array of scalar attribute names; [] reads only the ID. The full signed 64-bit contact ID is preserved without rounding. Attribute-definition reads require all_contact_read and have no contact ID field. Both operations require an unchecked personal-data acknowledgement and separate confirmation. Missing selected attributes remain absent and are listed separately; null, false, zero and empty strings remain distinct. Each invocation makes one attempt despite Retry On Fail. Save sends no request and Test connection is unavailable. Generic family credentials, OAuth, writes, automatic pagination and import aliases are unsupported. Real provider, Review, organization and durable behavior remain untested.

E-goi

E-goi setup and limits covers two native reads with egoiApi: one page of mailing-list metadata and one contact by exact list/contact IDs. List IDs remain decimal strings through the signed-int64 maximum; contact IDs are exactly ten hexadecimal characters with case preserved. Optional list limit and offset are numbers and support typed numeric expressions. Both operations start with the personal-data switch unchecked and require independent confirmation. Each read makes one attempt despite Retry On Fail. Missing optional values stay absent; null and empty strings remain distinct. Save sends no provider request and Test connection is unavailable. OAuth, sends, writes, email search, automatic pagination and import aliases are unsupported. Real provider, Review, organization and durable behavior remain untested.

Elastic Security

Elastic Security setup and limits covers two native v9 Kibana case reads with elasticSecurityApi. Use raw JSON with apiKey, baseOrigin, spaceId and apiVersion; JSON null selects the default space. List requires explicit numeric page and perPage, while Get requires a lowercase case UUID. Both begin with privacyAcknowledged unchecked and require separate confirmation. Private content reaches the bounded parser before metadata projection; Get checks owner after receipt. Optional counts preserve absence and permitted nulls; original canonical integer tokens prevent rounding. Each read makes one attempt despite Retry On Fail. Save sends no provider request; Test, writes, triggers, automatic pagination and import aliases are unavailable. Real provider, Review, organization and durable behavior remain untested.

Emelia

Emelia setup and limits covers three native REST reads with emeliaApi: campaign metadata, one campaign by a 24-character hexadecimal string ID, and one explicit list-metadata page. Lists require numeric page and limit; campaigns have no server pagination and reject more than 100 records locally. Personal-data acknowledgment starts unchecked, unchecking persists false, and each invocation requires separate confirmation. Full private response content reaches the bounded parser before scalar projection; missing fields, supported nulls, false, zero and empty strings remain distinct. Original numeric tokens are validated before conversion. Each read makes one attempt despite Retry On Fail. Save sends no request; Test, writes, discovery, automatic pagination and legacy GraphQL import aliases are unavailable. Real provider, Review, organization and durable behavior remain untested.

ERPNext

ERPNext setup and limits covers two native v1 metadata reads with erpNextApi and raw cloud/self-hosted JSON credentials. List reads one explicit page with numeric limit 1–100/default10 and offset 0–1,000,000/default0. Get requires an exact supported string DocType and document name; leading zeros and case remain intact. Privacy acknowledgment starts unchecked, can be unchecked again and remains separate from confirmation. Get receives the bounded full private document, potentially including personal data and child tables, before projection to name and supported creation, modified and docstatus fields. Names can themselves contain personal data. Missing, null, empty text and zero remain distinct; docstatus validates original integer tokens0/1/2. Every invocation makes one attempt despite Retry On Fail. Save sends no provider request; Test, discovery, triggers, automatic pagination and n8n import aliases are unavailable. Real provider, Review, organization and durable behavior remain untested.

ERPNext also provides document-create, document-update and document-delete. Create/Update take bounded fieldsJson with flat string-valued fields; the server validates required fields. Both privacy and change acknowledgments begin unchecked and remain separate from run confirmation. Changes may trigger business hooks or notifications; deletion may be permanent. Create/Update require a supported HTTP 200 document; Delete requires HTTP 202 with data: "ok" and returns only an acceptance acknowledgment. This follows the pinned Frappe v15.123.1 contract. A failed or unsupported response after dispatch leaves the effect uncertain, with no automatic retry, pre-read or verification call. See the document-change limits. Hosted approval, real provider, organization and durable behavior remain untested.

Freshworks CRM

Freshworks CRM setup and limits describes 35 native operations with freshworksCrmApi: six selected-record gets, six explicit page lists, search and contact lookup, and seven resource families with create/update/delete. Enter raw JSON with exactly apiKey and one lowercase tenant domain label on myfreshworks.com. Save sends no provider request, leaves authentication unverified and offers no Test connection. Select the native account and enable workspace live execution. Exact positive decimal string IDs retain precision through signed-int64 maximum; deal amounts remain decimal strings and whole expressions retain their types. Personal-data and write-effect switches start unchecked, can be unchecked again and do not replace separate invocation confirmation. Full private responses reach bounded parsing before selected scalar output. Retry On Fail never repeats these operations. Writes can trigger automation, notifications or deletion; an accepted response does not prove durability, and failures can leave an indeterminate change. No automatic pages, verification reads, import aliases or OAuth are added. Default durable approved-run confirmation remains Google-only; local callback tests do not prove human consent. Real provider, hosted Review, organization and durable behavior remain untested.

Ghost post operations

Ghost setup and limits describes two native credential profiles and seven bounded operations: Content and Admin Get/List, plus Admin Create/Update/Delete. Use exact saved ghostContentApi or ghostAdminApi accounts with raw adminOrigin/apiKey JSON. All personal-data acknowledgments and the three write-effect acknowledgments start unchecked; separate confirmation is required. Explicit numeric pages and string IDs retain their types. Empty HTML is preserved, omitted Update HTML stays absent and featured:false remains false; undefined is refused. Admin updates require a reviewed updated_at timestamp. Save sends no request or JWT and Test is unsupported. Publishing, scheduling and deletion can affect content/mail/automation; version negotiation itself can notify the owner. Each invocation makes one attempt despite Retry On Fail, and an uncertain write requires inspection before a manual retry. No import aliases, loaders, triggers or automatic pages are provided. Durable approved-run confirmation remains Google-specific; explicit live single-step authorization is version/segment bound. Local synthetic tests do not verify real provider, Review, organization, durability or human consent.

Gotify messages

Gotify setup and limits describes three bounded operations with selected gotifyApi raw JSON accounts: create one notification with an application token, delete one message or read one metadata page with a client token. Both roles remain distinct when supplied. Public root HTTPS origins only; Save sends no provider request and Test is unavailable. Privacy begins unchecked for every operation; writes also require an unchecked effect acknowledgment and independent invocation confirmation. Empty title and priority0 remain values; exact decimal string IDs/cursors and signed-int64 priorities retain precision. Whole text expressions must return strings. Full private messages reach parsing before metadata-only projection. One attempt applies despite Retry On Fail, without redirects, automatic pages or verification. DELETE200 is local acceptance, not durability proof; inspect uncertain writes before manual retry. No n8n import aliases or automatic parameter conversion are supplied. Other39 of42 documented API operations remain backlog. Real provider, Review, organization, persistence and human approval remain unverified.

Hunter email operations

Hunter setup and limits covers one domain-search page, finding an email and verification. All require personal-data acknowledgment and separate invocation confirmation. Finding and verification additionally acknowledge credit use and automatic lead changes; their GET method does not make them read-only. Pending verification returns without polling. No automatic retries or n8n import conversion are added.

Home Assistant reads

Home Assistant setup and limits covers Get state, List states, Get configuration, List event listeners and List service names. Select an existing native account and enable workspace live execution. Review the initially unchecked personal-data switch and confirm each read independently. Only selected scalars are returned; household attributes, coordinates and descriptions are omitted. One attempt is made even with Retry On Fail; device actions, writes, automatic pages and n8n import conversion are outside this profile.

Humantic AI profiles

Humantic AI setup and limits covers Create profile request and Get profile metadata. Select the exact native account, enable workspace live execution and review both initially unchecked acknowledgments before separately confirming. Creation is a write that starts processing and may enrich email profiles; metadata retrieval exposes selected status/ID/confidence fields without personality assessments. Persona accepts a JSON array of unique sales/hiring values, including an explicit empty array. Whole expressions retain types. One attempt applies despite Retry On Fail, without redirects, polling, updates or n8n conversion.

Iterable events, profiles and lists

Select the saved iterableApi account and enable workspace live execution. Event tracking, upsert, selected user fields, delete and list subscribe/unsubscribe require personal-data acknowledgment and separate invocation confirmation; writes also need effects acknowledgment. Global channel unsubscription needs its own acknowledgment. Use explicit typed JSON arrays and decimal-string IDs to preserve large values. Whole expressions retain types; one attempt applies despite Retry On Fail. See setup, precision and limits.

HighLevel legacy CRM and calendar actions

Select the exact native PIT/location/version account and enable workspace live execution. Seventeen bounded contact, task, opportunity and calendar actions use initially unchecked personal-data and applicable effects/upsert acknowledgments plus independent confirmation. Whole expressions preserve types; optional empty contact names remain values and nullable names require a typed null expression rather than literal textbox text. Lists are bounded partial responses. See setup, supported fields and limits.

Keap CRM actions

Choose the saved native keapApi account and one of29 actions: twelve reads and seventeen writes. Enter exact decimal-string IDs; list limits are numeric1–20, with an explicit operation-specific filter. Lists return one partial first page. Personal-data and applicable write/effect acknowledgments start unchecked and require separate confirmation. Order prices stay original decimal strings; line items and ID arrays use strict bounded fields. Upsert can merge contacts; email transport acceptance is not delivery. Inspect indeterminate writes before manual retry. File metadata has a descriptor key, not a fabricated file ID; note text can contain personal data. Whole expressions preserve original scalar types. See supported fields and limits.

Invoice Ninja hosted-v5 blocks

Use the selected invoiceNinjaV5Api account for 31 bounded operations across clients, invoices, tasks, payments, expenses, bank transactions and quotes. IDs and financial decimals are exact strings; list page/perPage values are numbers. Invoice/quote line items require a strict array, and edited empty notes or descriptions stay empty. Privacy and applicable effect switches start unchecked and require separate confirmation. Email requests do not prove delivery. Native queued segments refuse duplicate or uncertain replay, including preparation failures that may have sent no HTTP. See fields and local limits.

LinkedIn publication blocks

Choose one of three actions: Create text post, Create article post or Create post with existing image. Use the selected linkedInAccessTokenApi account, an explicit person/organization author URN, visibility and literal text. Article metadata and an existing thumbnail URN are required; image posts use an existing image URN. Review personal data and exposure, check the initially unchecked publication acknowledgment and confirm separately. Whole expressions retain original scalar types. Results contain only the exact header URN; there is no upload, readback or visibility verification. See fields and limits.

Mandrill Transactional send blocks

Choose Send HTML message or Send template message with the selected mandrillTransactionalV14Api account. Enter the explicit sender, 1–10 unique recipients, subject and async boolean; review privacy, email and billing effects and confirm separately. HTML offers optional text: untouched stays absent and explicitly cleared text stays empty. Templates require the existing slug, two explicit strict variable arrays and merge-language choice. Whole expressions retain original types, including false and empty string. Returned recipient statuses/actual IDs do not prove delivery. See fields and local limits.

LingvaNex and Magento native blocks

LingvaNex offers bounded language listing and literal translation with explicit from/to codes. Select lingvaNexCloudV3Api, enable its workspace control, review usage or text disclosure/charges, check the initially unchecked acknowledgments and confirm separately. Magento offers customer/order/product get/list plus nine explicit effects using saved magento2Api. Review privacy and all operation-specific account/email/overwrite/deletion/financial/invoicing/cancellation/fulfillment/inventory acknowledgments. Lists are one explicit page; numeric IDs, price and quantity remain original strings before controlled numeric wire serialization, with no float/storage guarantee. Required false options and item arrays preserve types through whole expressions. Neither provider follows returned URLs or retries automatically. See LingvaNex fields and Magento fields.

Household, messaging, relationship and invoice blocks

Home Assistant actions9, Matrix9, Mailcheck email-check, Monica25, Mindee InvoiceV4 and MessageBird SMS/balance use explicitly selected native accounts and workspace live controls. All applicable privacy/effect policies start unchecked and remain editable; confirmation is independent. Whole expressions preserve original scalar/object types. Store business booleans false explicitly; use {{ null }} for nullable fields, and keep empty description/cursor text distinct. Monica reminder replacement requires NONNULL contactIdFilter and explicit delible; scheduling can affect EVERY account user and replace outboxes. Camera stores an owned image; Mindee uploads one admitted owned snapshot and may buffer a larger file before refusal. Lists return one bounded response. See all47 finite operations and limitations.

Receipt, Calls and Mocean blocks

Choose ReceiptV5 with its separate mindeeReceiptApi account and one owned file, Monica Calls with the existing Classic account, or Mocean SMS/voice with its two-secret account. Receipt accounts cannot run Invoice actions or vice versa. Calls create/replace require Gregorian calledAt, nullable content and nullable contactCalled; false/empty remain explicit and null uses {{ null }}. Replacement has no contact selector/filter/reassignment. Mocean sends to one digits-only recipient with explicit message/sender/language where required. Policies begin unchecked and confirmation is separate. See all eight actions.

Mistral OCR and Teams reads blocks

Mistral document-extract-text selects one owned PDF fileId and requires document-disclosure/paid-processing acknowledgments plus independent confirmation. The first eight requested pages yield selected extractedText/pageCount only; returned count is not the complete document count. Teams offers eight finite Graph v1 reads with exact string IDs, data-access acknowledgment and separate confirmation. Message pages require numeric pageSize1–25. First-page continuation is required null via {{ null }}, not ordinary text null; explicitly supply a supported nextPageUrl for another page. HTML remains opaque text. See all nine field and output contracts.

Monica Activities and Teams mutation blocks

Activities create/replace require an explicit participant set, Gregorian happenedAt, required nullable activityTypeId/description and nonblank summary; replacement overwrites all five fields and participants. Preserve empty description/string0 versus null with {{ null }}. Teams mutations use manually supplied IDs; Planner update/delete require an explicit quoted ETag. Planner numeric0 remains distinct from string0/null. All applicable privacy/change/overwrite/deletion controls start unchecked and need separate confirmation. HTTP204 outputs only httpStatus, not verified remote state. See the fourteen field and response contracts.

Agent prompt and public synonym blocks

Perplexity Ask Agent (fast) accepts prompt, acknowledgePersonalData and acknowledgePaidRequest. OpenThesaurus Look up German synonyms accepts word, actual boolean baseform and acknowledgeQueryDisclosure. All acknowledgment switches start unchecked; typed expressions preserve booleans and independent confirmation is required. Agent output is local answer aggregation plus unnumbered selected sources, without citation mapping. Synonym output retains actual attribution and displays its source link without nofollow plus actual copyright/license. No automatic imported Sonar conversion or dummy OpenThesaurus account is created. See the two field, response and privacy contracts.

Outlook reads and Monica conversation/contact blocks

Outlook requires a manually supplied lowercase mailbox UUID, exact resource IDs, acknowledgeMailboxAccess and, for pages, numeric pageSize1–25 plus required nullable continuationUrl. Use {{ null }} for the first page. Monica conversation dates use UTC timestamps; contact dates use calendar days. Contact create/replace require every profile/date field, including explicit nullable values and booleans; replacement clears/rebuilds date/reminder state. Every applicable acknowledgment starts unchecked and needs actual true plus independent confirmation. Contact get can update view state. No automatic ID lookup, cursor follow, retry or full-data guarantee is added. See the23 field and response contracts.

Outlook write blocks

The native Graph token profile has28 operations, including calendar/contact/folder/event create/update/delete and draft create/update/delete/send. Enter mailbox/resource IDs explicitly. Required nullable selections need typed {{ null }}; in draft updates null omits a field and empty text clears it. Draft subject/bodyText/toName preserve intentional whitespace; other fields retain existing normalization. Switches start unchecked and remain independent from execution confirmation. Send includes stored undisplayed To/Cc/Bcc/content/attachments; deletion can affect a non-draft message, and event writes can notify attendees. See the sixteen finite contracts.

urlscan and Marketstack blocks

urlscan Submit URL with private visibility needs an HTTPS URL and actual true disclosure acknowledgment, initially unchecked, plus separate confirmation. Get one selected scan result takes a manual UUID and has no acknowledgment or confirmation requirement. Marketstack ticker.get/exchange.get/eod.latest need a selected native v2 account and initially unchecked query/quota acknowledgment plus separate confirmation. Supply one uppercase symbol or four-character MIC, and explicit numeric EOD limit1–25/offset0–1000000. One attempt/page only; numeric EOD observations retain original JSON text. See fields, privacy and response limits.

npm public metadata blocks

Choose one of four npm public registry reads. Version lookup needs packageName/version; versions listing needs packageName/explicit numeric limit1–25; dist-tags needs packageName; search needs query/limit1–25/offset0–1000, preserving offset0. No account or acknowledgment field is added. Review the separate execution confirmation and selected output, not raw package scripts or people. See the npm field, response, privacy and timing contract.

BigQuery and Business Profile metadata

BigQuery tables.get takes string projectId, datasetId, tableId; datasets.listPage takes string projectId, explicit numeric integer maxResults 1–100 and optional exact pageToken. Preserve table spaces, leading zeros and continuation whitespace. Business Profile locations.get takes exactly {} and uses the location saved in the selected account; locations.listPage uses its saved account destination with numeric integer pageSize 1–100 and optional exact pageToken. Whole expressions must preserve string/number types. Extra keys and per-call Business Profile destination overrides are refused. Select the exact native manual account and review independent execution confirmation; no acknowledgment input is added. Workspace booleans override environment defaults; absent flags enable and exact false disables when no override exists. One attempt/page, no automatic continuation, SQL/jobs/table data, reviews/posts or writes. Quoted BigQuery counters stay strings; Business Profile pages always report accountComplete:false. Ordinary configuration, history and downstream steps may retain inputs and metadata. Destination IDs do not establish ownership, grants or account/location association; project eligibility and real provider behavior remain unverified. See the full account, input and response contract.

Chat spaces, Storage metadata and plain-text sentiment

Chat spaces.get needs exact resourceName including spaces/; spaces.listPage needs numeric pageSize 1–100 and optional exact pageToken. Storage buckets.getMetadata needs bucket; objects.listPage adds numeric maxResults 1–100, optional exact Unicode prefix and pageToken. Preserve continuation whitespace. Both services require independent confirmation with no acknowledgment input. Sentiment analyzeSentiment needs exact text, two separate initially unchecked disclosure/billing choices and independent confirmation. Markup stays literal. Select the matching native manual account and live control. One attempt, no automatic pages or content downloads; ordinary configuration, results and history may retain submitted text and metadata. Selected Storage integers and sentiment numeric tokens remain strings. See the full contract.

Translate text and read selected contacts

Google Translate translate requires literal text, targetLanguage and two initially unchecked boolean choices, acknowledgeTextDisclosure and acknowledgePossibleBilling, plus independent execution confirmation. Google Contacts contacts.get requires the exact resourceName including people/ and selected personFields; contacts.listPage requires selected personFields and explicit pageSize 1–100, with an optional exact pageToken. Select names, emailAddresses, phoneNumbers, organizations or birthdays. In the parameter controls, choose fields from the multiselect; a whole expression must resolve to the typed array. Contacts has independent confirmation and no acknowledgment input. Preserve opaque IDs and continuation whitespace. Select the matching manual account and enable its workspace control. One attempt, no automatic pages or token refresh. Inputs and results can enter retained workflow history; inspect an uncertain translation before repeating it. See the full account and response contract.

OneDrive and SharePoint selected metadata blocks

Renaming one of these blocks keeps its selected service, operation and account. Change the block's configuration to select a different account or operation.

OneDrive items.getMetadata requires exact string itemId; items.childrenPage additionally requires numeric integer pageSize 1–100. SharePoint site.get accepts exactly {} and uses the site saved in the selected account; lists.firstPage accepts only numeric integer pageSize 1–100. Select the matching manual credential; no drive/site override or acknowledgment input is accepted. Review independent confirmation. Workspace booleans override environment defaults; otherwise only exact false disables and an absent flag enables. OneDrive size is original Int64 decimal text. Supported omission, null, false, order and duplicates stay distinct. Children returns items, pageScope:"first-page", accountComplete:false, hasNextPage; lists returns value, hasNext, pageScope:"first-page". Neither returns or follows nextLink, accepts continuation, establishes completeness or includes content, downloads, writes, sharing or search. Inputs and results may persist in history. SharePoint site.get is a new native Graph read; shipped REST action/field parity and import aliases are unavailable. See the full account, input and response contract.

Google Slides selected metadata reads

Choose Google Slides · selected metadata reads (google-slides-metadata) for this separate manual-token profile. The existing Google Slides OAuth/service-account integration and its eight actions stay separate. Save googleSlidesManualBearerApi with only accessToken, then select that exact saved account on the block. Supply an already-acquired bearer token without a Bearer prefix. Save is local and unverified with zero provider requests; Test, token acquisition, refresh and service-account signing are unavailable. Saving establishes neither account access, scopes, expiry nor execution authority. Google quota applies.

Select a nonempty unique array from the operation's choices. Whole expressions must produce actual string arrays, not JSON text. IDs use the native subset of 1–256 ASCII letters, digits, underscores or hyphens; preserve case and leading zeros. URLs are refused. These are local support limits, not Google's complete ID grammar. Returned identity must match the exact requested ID.

The response includes the identity and only requested metadata. slides permits at most 100 summaries; a larger response is refused rather than truncated. Each summary contains objectId and supported pageType, revisionId and slideProperties; the latter projects only layoutObjectId, masterObjectId and isSkipped. Missing fields, null and false remain distinct where supported. No full slide content, notes, comments, thumbnails, media downloads, writes or automatic follow-up is included.

Independent runtime confirmation remains required, without an extra acknowledgment input. WORKFLOW_LIVE_GOOGLE_SLIDES_METADATA_MANUAL_BEARER controls this provider: an explicit workspace boolean overrides the environment; otherwise exact false disables and an absent flag enables. IDs go to Google, and inputs, selected results and history may persist normally and reach downstream steps. The two reads provide partial service coverage, without n8n field/item/error parity, import aliases or credential conversion. Local synthetic checks do not verify real Google access, quota, human consent or Review behavior.

Microsoft Entra ID selected metadata reads

Choose Microsoft Entra ID · selected metadata reads (microsoft-entra). Save microsoftEntraManualBearerApi with only an existing accessToken, without a Bearer prefix, then select that exact saved account on the matching block. Save is local and unverified with zero provider requests; Test is unavailable. Saving does not establish identity, permissions, audience, expiry, tenant consent, privacy consent or execution authority. Token acquisition, refresh, OAuth consent and n8n credential conversion are unavailable.

Get accepts only a GUID-shaped existing ID in this native subset: hexadecimal 8-4-4-4-12 groups separated by hyphens. Preserve exact case and leading zeros. Microsoft Graph IDs are opaque, not universally GUIDs; UPNs, email addresses, URLs and other ID shapes are outside this input profile. A differently serialized response ID is refused, without normalization. Numeric strings are not page sizes. Extra input keys and arbitrary metadata fields are refused.

The page is always an incomplete inventory. Continuation is discarded, never returned or followed, and no later-page input is supported. hasNext:false does not prove complete directory coverage. Selected missing fields, null, false, empty strings, order and duplicates remain distinct where supported. There is no discovery, search, write, membership change or automatic paging.

Independent runtime confirmation remains required, without an extra acknowledgment input. WORKFLOW_LIVE_MICROSOFT_ENTRA_MANUAL_BEARER controls this provider: an explicit workspace boolean overrides the environment; otherwise exact false disables and an absent flag enables. Personal and directory metadata may persist in ordinary workflow inputs, results, history and downstream steps. Permissions, account eligibility and provider charges remain unverified. These four reads provide partial ordinary/Tool service coverage, without n8n locator/field/item/error parity or import aliases. Static shipped-source evidence and local synthetic checks do not prove live Graph access, human consent or Review behavior.

Graph Security and Workspace Directory metadata blocks

Microsoft Graph Security · selected profile reads has Get selected security control-profile metadata (secureScoreControlProfiles.getMetadata, exact id) and Read one security control-profile first page (secureScoreControlProfiles.firstPage, numeric integer pageSize 1–100). Select the exact saved microsoftGraphSecurityManualBearerApi account containing only accessToken. IDs preserve case and zeros within [A-Za-z0-9_-]{1,256}; whitespace and URLs are refused. Full upstream profiles reach the transport before local projection; the decoded-body cap is 1 MiB, not raw-wire size. No $select, scores, alerts, comments, content, tenant URLs or writes.

Google Workspace Directory · metadata reads has Get user metadata (users.getMetadata, exact userId), Read first user metadata page (users.firstPage, pageSize), Get group metadata (groups.getMetadata, exact groupId) and Read first group metadata page (groups.firstPage, pageSize). Page size is an explicit numeric integer 1–100. Select the exact saved googleWorkspaceDirectoryManualBearerApi account with accessToken and required customerId outside operation input. IDs use [A-Za-z0-9_-]{1,128}, preserving case and leading zeros; padded text is preserved raw and refused, never normalized. No email/domain/URL/my_customer customer alias. Returned users must match the saved customer; group responses have no customer field, so group-customer ownership cannot be independently verified. directMembersCount stays quoted 1–19 digits at most 9223372036854775807, including zero and leading zeros.

All first-page outputs always have inventoryComplete:false and only hasNext, even for empty pages or hasNext:false. No continuation input, exposure or following is supported.

Save the exact matching manual account locally and unverified, then select that saved account on the matching block. Save account sends zero provider HTTP requests; Test connection is unavailable. Existing credential disclosure choices remain separate from independent per-run execution confirmation. Saving grants no execution authority and adds no operation acknowledgment field. Token identity, audience, expiry, permissions, customer eligibility, quota, licensing and any charges remain unverified. Inputs, projected outputs, downstream data and run history may persist in ordinary workflow storage. No OAuth acquisition, refresh, signing, discovery, search, writes, membership changes or n8n import aliases are provided.

The separate live controls are WORKFLOW_LIVE_MICROSOFT_GRAPH_SECURITY_MANUAL_BEARER and WORKFLOW_LIVE_GOOGLE_WORKSPACE_DIRECTORY_MANUAL_BEARER; explicit workspace booleans override the environment, otherwise only exact false disables and absent enables. These are selected manual reads, without original n8n node or security-score parity. See Graph inputs and projections, Directory inputs and projections and account fields.

Webex, SecurityScorecard and Wise metadata blocks

Select the exact saved account for Webex · selected meeting metadata (webexMeetingManualBearerApi, only accessToken), SecurityScorecard · selected company reads (securityScorecardReadApi, only apiKey) or Wise · profile metadata reads (wiseProfileReadApi, only apiToken). Save is local/unverified with zero provider HTTP and Test is unavailable. Existing eligible accounts and permissions are prerequisites, not verified facts. Saving grants no execution authority; credential disclosures and independent per-run confirmation are separate, without an operation acknowledgment input.

Webex IDs are [A-Za-z0-9_-]{1,512}, preserving case and zeros; other opaque forms, URLs and whitespace are refused. Get leaves current=false and List leaves current=true/meetingSeries defaults. Series overlap the selected range; returned timestamps can lie outside it. continuationPresent only reports nonempty Link-header presence, not parsed next-page proof; no links are returned or followed.

SecurityScorecard refuses URLs, ports and custom IDs. The selected key must already access the scored company or portfolio; terminal 403/404 errors never create, enroll or upgrade. Optional summary score stays absent; factor issueCount is derived from bounded issue-summary counts, with absent versus empty arrays distinct and null selected values refused. See exact domain, factor and count bounds.

Wise fixes api.wise.com/2026Q4 with no legacy TransferWise v1 fallback. Supply a positive canonical int64 decimal string no greater than 9223372036854775807, preserving values above 2^53; zero, leading zeros and input numbers are refused. Original numeric response IDs become exact strings. The bare-array list preserves duplicates; optional currentState stays absent when omitted, while null is refused. No financial actions are offered.

Inputs, projections, ordinary history and downstream data can persist. Full bounded upstream bodies may contain private fields before projection. No automatic paging, retries or returned-URL following occurs. Each read makes one GET attempt under a 10-second provider deadline and a 262,144-byte decoded-body cap. Current rights, expiry, quota, eligibility and compatibility remain unknown. The separate live controls are WORKFLOW_LIVE_WEBEX_MEETING_MANUAL_BEARER, WORKFLOW_LIVE_SECURITYSCORECARD_READ and WORKFLOW_LIVE_WISE_PROFILE_READ: explicit workspace booleans override the environment, otherwise only exact false disables and unset enables; later selected-account/live checks still apply. Partial ordinary/Tool service coverage adds no n8n import aliases or full action/field/OAuth parity. See Webex, Wise and credential fields.

Travis CI and Tapfiliate selected metadata

Choose Travis CI · selected hosted build metadata or Tapfiliate · program affiliate metadata, the exact saved workspace account and one operation. The operation fields are:

Travis build IDs are canonical positive decimal text through 9007199254740991, without leading zeros, spaces, exponents or coercion. It requests fixed hosted .com v3 /build/{buildId} or /builds?limit={limit}&offset=0. The response must provide numeric IDs and @type:"build"/"builds"; get must match the requested ID. Selected id becomes exact decimal text. Optional number is 1–32 decimal digits, state matches [a-z][a-z0-9_ -]{0,63}, and private is boolean. Absent/null optional fields are omitted; invalid nonnull fields refuse the response. Lists preserve order and duplicates, without fetching later pages. Original get/getAll are narrow partial subsets; cancel, restart, trigger, include, sorting and full outputs remain unsupported.

Tapfiliate IDs use exact [A-Za-z0-9_-]{1,128} text, without trimming or normalization. Get calls /1.6/programs/{programId}/affiliates/{affiliateId}/; list calls exactly /1.6/programs/{programId}/affiliates/?page=1 on fixed api.tapfiliate.com. The documented service page default is 25, with a local cap of 100; no invented page-size option is available. Results require id, omit absent approved, preserve false and refuse null or nonboolean approval. Affiliate-ID matching does not independently verify program ownership. Order and duplicates are preserved. Only analogous program-affiliate metadata subsets are provided; the full original 12 actions, outputs, filters, writes and pagination remain incomplete.

Get returns {"item":{…},"inventoryComplete":false}; list returns {"items":[…],"inventoryComplete":false}. Both require independent per-run confirmation, separate from existing credential disclosures; there is no extra acknowledgment or billing field. Save sends zero provider HTTP, remains unverified and has no Test. Use workspace credentials for ordinary workflows; organization credentials are unsupported. Direct workspace membership, organization membership and execute authority are required; organization visibility alone is insufficient.

Each operation attempts one GET even with workflow maxTries:3, including after 429 or a transport failure. The total deadline is 10 seconds including preparation and projection; limits are 262144 decoded exposed body bytes, 8192 reader calls, JSON depth 16 and 10000 nodes. These do not establish raw-wire or decompression confinement. The full bounded private body enters memory before projection; inputs, results, history and downstream data use ordinary workflow retention. No retries, polling, followed URLs, full inventory, original action/import aliases or Tool wrappers. Provider access, quota, rights, expiry, eligibility and charges remain unverified. See credential fields and live controls.

Onfleet, Storyblok and PostHog selected metadata

Select the exact saved account for the named native provider. Every operation requires independent per-run confirmation, separate from credential disclosures. Saving fields, saving/reopening an unverified account or declining confirmation sends no provider request; Test is unavailable. There is no extra acknowledgment or billing-choice input.

Onfleet teamId is exactly 24 characters from [A-Za-z0-9_*~-], preserving case; get requires the response ID to match. Snapshot takes no fields, refuses more than 100 entries and duplicate IDs, and never promises full inventory. Absent metadata stays omitted; null stays null. Non-null millisecond timestamps are original nonnegative integer decimal text through 9007199254740991, without rounding. No tasks, drivers, routing or writes.

Storyblok storyId and PostHog projectId are positive canonical decimal text through 9007199254740991: no zero, leading zeros, whitespace, numbers, signs, UUIDs, URLs or slug aliases. String ID expressions preserve type and must return strings. Enter optional limit as a numeric integer 1–25, default 25; these editor number controls do not offer whole expressions. Extra fields are refused.

Storyblok uses only EU CDN v2 published stories, page 1. Each story returns id, lowercase uuid, name, createdAt, required nullable publishedAt and isStartpage, with optional nullable updatedAt. The supported RFC3339 subset uses valid dates in years 0001–9999, seconds, up to nine fractional digits and Z/numeric offset, excluding leap seconds and -00:00. No content, slugs, URLs, management writes or other regions. Duplicate list IDs or UUIDs refuse the result.

PostHog uses explicit saved region and organization outside operation inputs, offset 0, and exact get-ID/organization matching. Six locally required non-null fields survive: id, organizationId, uuid, name up to 200 text units, isDemo and timezone up to 64. False is retained; timezone text is not certified as IANA. Duplicate list IDs refuse. Own personal automation requires existing project:read, actual membership/visibility and organization policy; project-restricted keys are unsupported for listing. All five original n8n PostHog write actions remain unported.

Use workspace credentials for ordinary workflows. Organization-scoped credentials are unsupported. Organization-visible workspaces still require direct workspace membership, organization membership and execute authority. Full bounded upstream private fields enter memory before projection, including content and tokens/settings. Ordinary input/output/history/downstream retention applies. A single GET attempt may consume quota, with no retry, paging or followed URLs and no charge/rights guarantee. These are partial native service reads without old action/import aliases, Tool wrappers or full n8n compatibility. See accounts and live controls.

Paddle Billing metadata and Plivo SMS

Select the exact saved workspace account and an operation. Text expressions must return exact strings; perPage is numeric. No arbitrary URL, authentication header or request-body override is available.

Paddle IDs match pro_[a-z0-9]{26} exactly. Products select only productId, nonempty name up to 200 characters, status active/archived and type standard/custom. Get must match the requested ID. List requests fix order_by=id[ASC]; returned order and duplicates are preserved; local default 20/cap 50 differs from vendor default 50/max 200/descending order. Order and duplicates are preserved rather than locally sorted. Every list response needs a nonempty pagination.next string even when has_more:false. It must match the fixed HTTPS origin, products path, filters and cursor policy. It is never followed or returned as a URL. When hasMore:false, nextAfter is null; true requires nonempty items and a progressing cursor relative to any supplied cursor. Copy nextAfter to after only for a separately confirmed manual run. Inventories always remain incomplete.

Plivo to is one plus-prefixed number with 2–14 digits, beginning with a nonzero digit. from uses that numeric subset or 1–11 ASCII letters/digits including at least one letter. text must be nonblank, at most 737 UTF-16 units and 2948 UTF-8 bytes; no unpaired surrogates or C0 controls except tab, line feed and carriage return. Valid whitespace is preserved. These conservative local limits do not prove vendor acceptance, sender eligibility, consent, segment count or charges. A response requires the exact queue message, one lowercase canonical message UUID and a canonical API UUID. Unknown response fields, including original number lexemes, are scanned before selected projection.

Paddle perPage uses a numeric control and status a select; after and productId are text controls that accept expressions. Plivo from and to are text controls; text is a textarea, and all three accept expressions. To preserve explicit carriage returns in text, use a whole expression returning the exact string. Literal carriage returns in an HTML textarea normalize to line feeds.

Each operation makes one request attempt without redirects, retries, polling or automatic continuation. Paddle uses one GET with a 10-second total budget including preparation and projection. Plivo uses one POST with a 10-second budget after preparation and an 11-second gateway timeout. Both bound decoded exposed response bytes to 262144, reader calls to 8192, JSON depth to 16 and nodes to 10000; overflow is refused, not truncated. Plivo selected output is bounded to 4096 bytes. These are not raw-wire or decompression confinement claims. A post-invocation Plivo error remains uncertain; abort cannot cancel a remote effect.

Use an existing workspace credential and select that exact account. Organization-scoped credentials are unsupported. Execution requires direct workspace membership, organization membership and execute authority; organization visibility alone is insufficient. Credential disclosure is separate from required independent confirmation for each run.

The full bounded upstream body enters server memory before projection. Inputs, selected results, workflow configuration, history and downstream data may be retained. Rights, permissions, expiry, account eligibility, quota and charges remain unverified. See accounts and run rules.

ProfitWell Metrics and Peekalink Preview

Select the exact saved workspace account and independently confirm each operation.

ProfitWell settings require strings locally bounded to id 256, name 1024, timezone 128 and currency 16 characters. Current primary prose names these fields without a response schema/example; requiredness, top-level shape, types and bounds are local policy, not vendor promises. Daily selects from fifteen documented metrics, with per-metric plan eligibility; the month must be the current or previous UTC month and is checked again at dispatch. Each series validates actual unique calendar dates, at most 31 points, independently and sorts chronologically rather than aligning by array position. Monthly selects from forty documented metrics with separate plan eligibility; validate the entire requested history before sorting and choosing its newest suffix. Missing months stay missing, invalid/duplicate dates refuse and nullable monthly rates survive. No server history limit, paging or start/end query is implied. Plans reads documented active-customer plan IDs ordered by MRR, with local string length 1–128 and duplicates preserved. The documented server maximum/default is 150; this block defaults to 25. There are no hidden options requests.

Peekalink URL validation rejects userinfo, fragments, IP literals, local/reserved hostnames, controls, explicit ports and known credential query/path forms. Unknown arbitrary secrets cannot all be discovered. DNS and provider redirect safety are unverified. ID preserves the original numeric spelling as text. Status 100–599 and title/description bounds 1024/4096 are local policy; type and updatedAt follow the selected current primary enum/date contract. Optional title and description may be absent. Preview is external processing with fetch/cache/media storage and possible charges. HTTP 200/ok:false and every post-invocation failure are sanitized uncertain outcomes; do not automatically retry. No image/platform/cookie/source/snapshot fetching, raw output or availability substitute is provided.

All five native operations have a 10-second total budget beginning before credential preparation, excluding outer database and confirmation latency. Bounds are 262144 decoded exposed response bytes, 8192 reader iterations, JSON depth 16 and 10000 nodes, fatal UTF-8 and duplicate rejection; overflow refuses instead of truncating. These do not bound raw wire bytes, decompression or operating-system execution. Full-body scanning includes ignored keys, strings and original numeric spellings, with supported exact encoding variants; arbitrary mixed/nested/encrypted secrets cannot all be recognized.

Use an existing workspace credential and select that exact saved account. Organization-scoped credentials are unsupported. Execution requires direct workspace membership, organization membership and execute authority; organization visibility alone is insufficient. Credential disclosure remains separate from independent confirmation of every run. Saving and reopening send zero provider HTTP, leave the account unverified and grant no execution authority; Test is unsupported.

Full bounded upstream bodies enter server memory before selected projection. Inputs, selected financial or preview results, workflow history and downstream data use ordinary retention. Rights, expiry, account eligibility, quota and charges remain unverified. These selected subsets do not provide full n8n action, field, item-loop, raw-output or error compatibility, import aliases or automatic Tool materialization.

The catalog labels are ProfitWell · selected metrics and Peekalink. Credential accessToken and apiKey controls are required passwords. Operation labels are Get selected company settings, Get selected daily metrics, Get selected monthly history, List active plan IDs and Preview one public link. Settings has no parameter controls. Daily month, optional planId and preview url use text controls with expressions. metrics uses multiselect checkboxes; its expression toggle accepts a JSON array or a whole expression returning a typed array. limitMonths and limit use numeric controls with an Expression toggle; expressions must return numbers. Defaults are applied when omitted; a newly rendered blank control is not proof that a default has already been stored.

Daily output is month, series:[{metric,points:[{date,value}]}], inventoryComplete:false and page:{mode:"selected-month",maxDays:31,fetchedRequests:1,nextCursor:null}. Monthly output is series, inventoryComplete:false and page:{mode:"local-history-suffix",limitMonths,upstreamPaginationSupported:false,nextCursor:null}. Original numeric values are strings; nullable monthly rates remain null. The plan list field is exactly plan_ids. The gateway allows 11 seconds around the native 10-second budget; this does not bound outer database or independent confirmation latency.

With planId, daily supports only churned_recurring_revenue, cumulative_net_new_mrr, cumulative_new_trialing_customers, future_churn_mrr, new_recurring_revenue, reactivated_recurring_revenue, recurring_revenue. Without it, all fifteen daily checkbox options are available.

With planId, monthly supports only active_customers, average_revenue_per_user, churned_customers, churned_recurring_revenue, churned_recurring_revenue_delinquent, existing_customers, existing_recurring_revenue, growth_rate, lifetime_value, new_customers, new_recurring_revenue, reactivated_customers, reactivated_recurring_revenue, recurring_revenue, customers_churn_rate, revenue_churn_delinquent_rate, revenue_churn_rate, revenue_retention_rate, plan_changed_recurring_revenue, plan_change_rate. Without planId, all forty monthly options are available except plan_changed_recurring_revenue and plan_change_rate, which require a plan. planId must match [A-Za-z0-9][A-Za-z0-9_.:-]{0,127} exactly.

Phantombuster agent metadata and explicit write requests

Select Phantombuster (phantombuster-agent-metadata), the exact saved workspace account and one of these literal operation labels. Independently confirm every operation.

agentId and optional prevContainerId use text controls with expressions, requiring exact positive canonical decimal strings from 1 through 9007199254740991. This is a local supported range, not a vendor maximum. Numeric controls (limit, fromOutputPos, prevRuntimeEventIndex, maxInstanceCount), boolean switches (saveArgument, manualLaunch) and the prevStatus select each offer an Expression toggle: the expression must return the required number, boolean or allowed string. Descriptor expression:false does not disable those toggles. argumentsJson and bonusArgumentJson are optional TEXT controls with expressions, not JSON-object controls; an expression must return an object encoded as a JSON string. Each string is bounded to 8192 code units and UTF-8 bytes, permits JSON whitespace and forwards original whitespace. The client never merges the two strings; the vendor documents bonus-argument merging, whose actual behavior remains untested.

fromOutputPos and prevRuntimeEventIndex require nonnegative safe integer numbers. prevStatus allows exactly starting, running, finished, unknown, launch error, never launched; current launch error differs from original lauch error. maxInstanceCount requires an integer from 1 through 9007199254740991; original zero is unsupported. Omitted launch booleans and count use vendor defaults: omission does not mean false. There is no extra acknowledgment or billing field.

Get metadata requires the selected id to match the input. Optional name/scriptId are locally capped at 512 characters; script accepts the selected coffee/js filename pattern, environment is staging/release and lastEndType is finished/killed/global timeout/org timeout/agent timeout/unknown/no log timeout. Optional selected fields may be absent or null. Output metadata requires status plus two booleans; selected numeric positions, time and progress preserve original number spellings as strings. It does not expose raw logs, runtime events, progress labels or arbitrary results, and does not implicitly resolve containers. Only list adds the incomplete-inventory/local-prefix fields; the whole bounded array is validated before prefix selection, without query limits or pages.

agent.launchRequest executes stored automation, potentially using saved platform credentials and causing scraping, external changes, quota use or charges. saveArgument:true can change saved launch defaults. agent.deleteRequest is destructive for the exact confirmed agentId; documented 400 missing-agent and 403 running-agent responses are failures. Each write makes one POST and returns only the LOCAL {requestedAgentId:<exact input>,httpStatus:200,outcomeVerified:false}. HTTP 200 does not prove launch, deletion, success or a container identity. The retained primary Swagger describes HTTP 200 without a write response-body schema; locally, an empty or bounded valid JSON body is fully scanned and discarded. Any failure after invocation leaves the remote outcome indeterminate. Do not automatically retry or read back; inspect Phantombuster separately before deciding what to do. Actual launch and delete were never performed during verification.

Choose Phantombuster (phantombuster-agent-metadata) and an existing workspace account of type phantombusterMetadataApi. Its required apiKey is a password control with exact visible ASCII text, 1–4096 characters; do not trim it. Optional vendorOrganizationId accepts 1–128 ASCII letters, digits, underscores or hyphens. This selects the vendor organization using X-Phantombuster-Org; it grants no Plaice organization or workspace authority. Requests use endpoint-specific X-Phantombuster-Key, without old Key-1/v1 fallback.

Saving and reopening are local, send zero provider HTTP and leave the key unverified. Test is unsupported. Select that exact saved workspace account for a run; organization-stored credentials are unsupported. Execution requires direct workspace membership, organization membership and execute authority. Current-account revision and exact prepared-request binding are checked again before dispatch. Target-bound personal execution has not been tested. Credential disclosure does not replace independent confirmation of each of the five operations; there is no extra acknowledgment or billing input.

Each native operation has a 10-second budget starting before preparation, excluding outer database and independent confirmation latency; its gateway allows 11 seconds. Limits are 262144 decoded exposed response bytes, 8192 reader iterations including completion, JSON depth 16 and 10000 nodes. Overflow refuses rather than truncates. These are not bounds on raw wire bytes, decompression, allocation or operating-system execution. One attempt is used, with no automatic retry, readback, fallback, polling or returned-link fetch.

Full bounded upstream bodies, including discarded arguments, proxy passwords, storage, manifests or logs, may enter server memory before projection. Ordinary caller inputs, selected results, execution history and downstream data may be retained; this is not universal input redaction. Supported finite secret encodings are checked across the response, including ignored fields; arbitrary nested, mixed or encrypted encodings cannot all be recognized. Permissions, key acceptance, quota, account scope, human confirmation and real-provider behavior remain unverified.

All five original intents have finite partial contracts. No raw-output compatibility, item loops, resolveData, returnAll, import aliases, loader or automatic Tool materialization/full-parity claim is made.

Pushcut and SIGNL4 finite actions

Choose Pushcut with a saved workspace account of type pushcutApi, containing only {apiKey}, or SIGNL4 with signl4Api, containing only {teamSecret}. These are existing manually supplied secrets. Pushcut uses the raw API-Key header, not a secret notification-webhook URL key. SIGNL4 requires only the final opaque webhook team-secret segment, never the full URL or REST API key.

Save and reopen send zero provider HTTP and leave the account unverified. Test is unsupported, and no hidden option loader checks the account. Select the exact saved workspace account and independently confirm every operation, including metadata reads. Saving or accepting credential disclosure grants no execution authority. There is no extra acknowledgment or billing control. Direct workspace membership, organization membership and execute authority are required; organization-stored credentials are unsupported, and target-bound personal execution is untested.

Devices has a multiselect descriptor but no static enum or option loader: the mounted field is a JSON-array text input with an Expression toggle, not a checkbox list.

Pushcut notification names are one encoded path segment, locally at most 256 code units and UTF-8 bytes, excluding control characters, slash, backslash, question mark, hash, percent and dot/dot-dot names. Devices allows at most 100 unique nonempty strings of at most 256 code units and bytes each. Omitted devices uses the vendor all-devices default. Explicit [] is sent unchanged and never guarantees no recipients. Optional input/text/title preserve empty strings and whitespace and are sent only when present; input/text are locally capped at 8192 and title at 1024 code units and UTF-8 bytes. There are no hidden defaults or substituted values. Reads validate the whole bounded array of at most 500 objects; they do not load options, paginate or promise complete inventory.

SIGNL4 message is required but may be empty when title is nonempty. Message/title/service/externalId have local maxima 8192/1024/256/512 code units and UTF-8 bytes. Optional empty title, service and externalId are accepted in the form but omitted from the wire. Service uses X-S4-Service, not old service; no X-S4-SourceSystem:n8n is added. Alerting Scenario permits exactly single_ack or multi_ack; omission leaves the team default. Filtering true sends X-S4-Filtering:true; false and absence omit it, preserving team suppression rather than forcing false. Both the select and switch expose an Expression toggle requiring an allowed string or boolean. Descriptor expression:false does not remove that mounted toggle. Text/textarea controls accept expressions; expressions must preserve the required value type.

Location accepts a typed object with exactly string latitude and longitude, through JSON text or a whole expression returning that object. Coordinates are canonical decimal strings, at most 32 code units/bytes, within latitude ±90 and longitude ±180; numbers and extra keys are refused. Binary Property accepts 1–128 ASCII letters, digits, underscores or hyphens and names the upstream persisted workspace-file descriptor; it does not accept bytes. Resolve externalId is a nonempty exact string up to 512 code units/bytes and is never replaced with eventId. No numeric parameter is introduced by these operations.

Every operation makes one attempt, without automatic retries, fallback, readback, polling or returned-link fetching. Pushcut send returns an HTTP 200/202 receipt, with id and, for 202, scheduleTimestamp. SIGNL4 returns HTTP 201 with an optional nullable eventId; even {} is a valid selected result. Both envelopes keep outcomeVerified:false. A receipt proves neither delivery nor action execution, scheduling, alert creation, closure, acknowledgement or affected count. Resolve uses the original external correlation ID (externalId, sent as X-S4-ExternalID), never the returned eventId, and can affect multiple alerts. After invocation, a failure leaves the remote outcome indeterminate; inspect the provider separately before deciding whether to repeat a request.

For a SIGNL4 attachment, first produce a persisted file in the current workspace and pass its file descriptor under the selected Binary Property. Only PNG, JPG, JPEG, BMP, GIF, MP3 and WAV are supported, with a local 1 MiB limit checked before reading or copying. The file must still be ready, with matching size and SHA256 before and after the bounded read. Arbitrary bytes, URLs, local paths, filenames and MIME types are not accepted. Signature checks validate local magic/MIME, not full media decoding.

A standalone workflow pauses at that same SIGNL4 node after resolving the current file. Open its attachment confirmation, inspect the exact account ID, binary property, format, size and SHA256, then choose Approve attachment or Reject attachment. The approval expires after five minutes. Only the original actor can decide that exact request; changed accounts, changed files, stale decisions and replayed approvals are refused. The current file, account and execution permission are checked again after approval, together with live and cancellation state. The ordinary manual-run confirmation cannot authorize an upstream file whose digest is not yet known. Approval details omit secrets, private tokens and file content; normal file retention still applies.

Standalone placement includes the selected branch of a condition. Attachments in loop bodies, parallel branches, nested composite descendants, sub-workflows and single-step runs are unsupported and fail before preparation, approval callback or provider HTTP. No success marker or earlier approval bypasses this restriction; continuation for those placements remains future work. Ordinary Agent and human approvals keep their existing behavior.

The native 10-second budget starts before preparation and excludes outer database and independent-confirmation latency. Response limits are 256 KiB of decoded exposed bytes, 8192 reader iterations including completion, JSON depth 16 and 10000 nodes; overflow refuses rather than truncates. These are not raw-wire, decompression, allocation or operating-system bounds.

Full bounded private provider responses enter server memory before scanning and selected-output projection. Ordinary inputs, selected results, execution history, downstream data and workspace files may be retained; there is no universal input redaction. Finite known secret encodings are checked, without a guarantee for arbitrary mixed, nested or encrypted encodings. Provider permissions, routing, quota, costs, delivery, resolve affected count and actual human/provider behavior remain unverified. Pushcut configured actions can invoke shortcuts, web requests, online actions and other systems; dynamic content requires Pro, without an upgrade or billing-consent control. SIGNL4 routing can incur SMS, voice or escalation charges.

All five original intents have finite partial contracts. Original multi-item loops, final-item-only raw output, continueOnFail, Tool/import/raw parity and the durable Pushcut trigger callback are not provided. No full-parity or provider-readiness claim follows from the inventory.

seven and MSG91 message requests

Use the exact saved workspace account for seven (sms77Api with apiKey) or MSG91 (existing template SMS) (msg91Api with authkey). Save is local, zero provider HTTP and unverified; Test is unsupported. Each operation independently requires confirmation and makes one attempt without retry, redirect, fallback, polling or readback, including when outer maxTries:3 is configured.

Text and textarea controls accept whole expressions preserving their required value types. The two MSG91 selects also expose an Expression toggle despite descriptor expression:false; expressions must return strings, not numbers. Recipients is an actual textarea, not a multiselect or comma-separated fallback. Untouched optional controls remain absent in saved config. There is no hidden option loader or implicit item loop.

seven fields and defaults

To is a comma-separated string of up to 100 nonempty submitted number/contact/group entries, at most 2048 code units and UTF-8 bytes, without tabs or line breaks. This is not a cap on actual recipients: groups may fan out. From, when supplied, is empty or at most 16 numeric or 11 ASCII alphanumeric characters; verified voice senders/booked numbers and carrier caller-ID rules still apply. SMS From omitted from config is sent as an empty string; voice options.from is sent only when supplied. Explicit empty strings remain valid where specified.

SMS Message is locally limited to 1520 Unicode scalars and 6080 code units/UTF-8 bytes. Voice text is limited to 8192 code units and UTF-8 bytes. The Options textarea accepts a JSON object or a whole expression returning that object. It is not a collection of separate date, number or switch controls:

{
  "delay": "2027-01-01T00:00:00Z",
  "foreign_id": "case_A",
  "flash": false,
  "label": "",
  "performance_tracking": false,
  "ttl": 2880
}

The six SMS keys are exactly delay, foreign_id, flash, label, performance_tracking and ttl. Booleans must be JSON booleans; explicit false is sent as string 0, true as string 1, and omission stays omitted. Foreign ID and label allow empty strings or ASCII letters/digits/._@-, with local lengths 64 and 100. TTL is a JSON integer in local range 1–10080 minutes; omission leaves the documented vendor default 2880, without injecting that default into config. The original UI used empty delay/IDs/label, false booleans, TTL 2880 and an empty options object; those per-option defaults are not automatically added here. Tracking enables link shortening and needs a custom domain configured in seven; encoding, segments and group fan-out affect charges.

Delay must be a valid whole-second timezone-bearing date such as 2027-01-01T00:00:00Z, explicitly converted to Unix seconds (1798761600 for that example). Fractional seconds, impossible dates, ambiguous local time and empty delay are refused. This intentionally differs from the original raw date forwarding. Omitting delay is valid; an empty string is not omission. Voice options supports only {"from":""} or an allowed sender, with no inserted default.

Voice supports plain text or the local closed grammar voice(name), prosody(rate), self-closing break(time or strength), p, s and say-as(interpret-as). For example: <voice name="en-gb-female"><p>Hello</p><break time="500ms"/></voice>. This is not general SSML: remote audio, DTMF callback markup, loops, call control and other XML/SSML are refused. Existing vendor account webhooks and recipient keypress events may still operate. Caller ID, voice availability and live acceptance remain unverified.

MSG91 explicit existing-template migration

The original action took free-form from, to and message; this operation requires a preexisting approved template and exact variable names. There is no free-form fallback, message-to-template conversion, creation or discovery. Template Id uses 1–128 ASCII letters, digits, underscores or hyphens. Enter Recipients as strict JSON:

[{ "mobiles": "4917612345678", "OrderCode": "SYNTHETIC_A" }]

The local bounds are 1–100 objects, each with mobiles as a 6–15 digit international string starting with a nonzero digit and 0–32 case-sensitive string variables. OrderCode and orderCode differ. Variable names start with an ASCII letter and contain at most 64 ASCII letters/digits/underscores; reserved prototype names are refused. Values preserve whitespace and empty strings, with a 2048 code-unit/UTF-8-byte limit each. Numbers and comma-separated recipient text are refused. The whole JSON request is locally limited to 32768 UTF-8 bytes.

Short Url permits string 0 (off) or 1 (on); untouched config omits the key and schema validation supplies the local string 0 default. Short Url Expiry is absent unless supplied as a positive decimal string from 1 to 604800 seconds, without leading zeros; explicit empty string is invalid. Real Time Response is absent unless supplied as string 1 only. Its select has one value and no default; numeric 1, string 0 and empty expression results are invalid. Clearing an optional literal field omits it at form conversion; a whole expression returning an empty string preserves that type and is then refused. The retained official example calls expiry/realTimeResponse optional while CLI metadata calls them required; local omission follows the example and actual account compatibility remains unknown. Neither realtime response nor HTTP 200 proves delivery. URL shortening may track links.

Confirmation, output and retained data

All three writes need independent confirmation of the exact request and selected current workspace account. Saving grants no authority. Direct workspace and organization membership, execute permission, account revision, prepared identity and current live/cancel state are checked before invocation. Organization-stored credentials are unsupported; target-bound personal execution is untested. WORKFLOW_LIVE_SEVEN and WORKFLOW_LIVE_MSG91 use explicit workspace booleans before environment values; otherwise exact false disables and unset enables. No extra acknowledgment or billing-consent control is introduced.

seven uses fixed-origin POST forms with raw X-Api-Key and Accept: application/json; MSG91 uses JSON POST flow with raw authkey header. seven string 100 plus validated receipts means accepted only; 101 may be partial. MSG91 HTTP 200 with type success yields only the selected request ID and acceptance envelope. Neither output certifies delivery, cost, recipient count or template approval. Every post-invocation failure is uncertain; do not automatically repeat. There is no delivery polling, hidden retry or readback.

The native 10-second budget starts before preparation, excluding outer database and independent-confirmation latency. Response bounds are 256 KiB decoded exposed bytes, 8192 reads including completion, JSON depth 16 and 10000 nodes; overflow refuses rather than truncates. These are not wire, decompression, allocation or operating-system limits. Full bounded private responses enter memory before scanning and projection. Billing, echoed content and recipient details are omitted from selected receipts, but ordinary inputs, selected outputs, history and downstream data may be retained. Known-secret scanning does not recognize every nested, mixed or encrypted encoding.

Static shipped intent, retained current documentation and local synthetic tests do not establish real-provider acceptance, delivery, rights, costs, Review behavior or independent human confirmation. All retained REA unknowns remain unresolved. The four ordinary/Tool inventory rows are partial service mappings with zero import aliases; they do not materialize Tool wrappers or provide original item-loop, raw-output, continueOnFail, legacy free-form or general-SSML parity.

Configure YOURLS and public QuickChart controls

Use an existing yourlsApi account with signature and public HTTPS installation base, including a subdirectory but excluding yourls-api.php. Save, Advanced and Format are local/unverified, zero HTTP and without Test. QuickChart has authentication:none and credentials:[]; its public card has no Connect account action. Every operation still needs independent confirmation.

YOURLS Shorten URL uses a Url text input and Additional Fields JSON-object textarea. Only keyword/title strings are accepted in that object. Empty keyword/title differ from omission; an omitted title may let the provider fetch the target title. Expand URL and Get URL statistics use a Short Url text input. Text expressions preserve native string values; optional literal controls left untouched remain absent. Returned URLs are metadata and never navigation actions.

QuickChart Items is a collection with Add item/Remove. Its nested Chart Type is actually a text input, even though the descriptor says select: the collection renderer uses input for known primitive kinds. Nested Data, Chart Options and Dataset Options are JSON textareas, without nested switches or dropdowns. Data accepts bounded numeric/null/point data; per-item Chart Options only accepts horizontal boolean. Dataset Options accepts label, backgroundColor, borderColor, fill and pointStyle. Collection edits parse object/array JSON; incomplete text remains a draft and fails schema validation when invalid.

Top-level Labels Mode is a select for manually/array; Labels Ui is a textarea for {"labelsValues":[{"label":"Q1"},{"label":"Q2"}]}. Labels Array is a text input for strict JSON or a complete expression returning an array. Output is a text input defaulting to data. Top-level Chart Options is one text input for a JSON object, not four format buttons, size controls or a horizontal switch. Supply width/height/devicePixelRatio/format/ horizontal keys or rely on validation defaults; an untouched required object still needs a value. Allowed formats are png/pdf/svg/webp, dimensions1–2000 and ratio1/2. The default type is bar, size500×300, ratio2, PNG, horizontal false and dataset label Chart.

Items and Labels Mode have Expression toggles based on rendered control type; Labels Mode's descriptor expression:false does not remove its toggle. Text/textarea fields accept expression text without separate toggles; their schema conversion preserves native object/array/string types. Labels Array and top-level Chart Options are text controls despite descriptor expression:false. Do not treat descriptor labels as proof of mounted control behavior.

{
  "items": [
    { "chartType": "bar", "data": [1, 2], "chartOptions": { "horizontal": true }, "datasetOptions": { "label": "A" } }
  ],
  "labelsMode": "array",
  "labelsArray": ["One", "Two"],
  "output": "chartFile",
  "chartOptions": { "format": "webp", "width": 500, "height": 300, "devicePixelRatio": 1, "horizontal": false }
}

This normalized batch preserves first labels/options/output and last effective type; it does not execute an n8n item loop. Five chart types (bar/doughnut/line/pie/polarArea) and four binary formats are supported locally. Functions/plugins/external URLs are refused. The fixed POST returns an attachment at binary.chartFile, with reference/MIME/format/size/SHA256 and outcomeVerified:false, never a URL-only or inline-HTML substitute.

Signed confirmation binds the exact node/provider/operation/version/public-auth mode and run/workspace/actor/workflow/version/material. Current execute, direct workspace AND organization membership, live/run identity and cancellation checks still apply. YOURLS also checks exact account/revision/HMAC/prepared identity. Each invocation makes one attempt even with maxTries3; no redirect, retry, fallback, polling or readback. Any post-invocation failure is uncertain. Full bounded bodies enter memory before projection; ordinary inputs, history, outputs, files and downstream data may persist. JSON256 KiB/image1 MiB,8192 reads,depth16/ 10000 nodes and native10s are local bounds, not wire/OS/allocation guarantees. Restricted binary profiles are not full viewer conformance. The write-only file lifecycle scans and adds typed run references, rolling back owned files on failure; no aggregate workspace quota is claimed. Local PostgreSQL tests cover current workspace/organization authority and revocation with synthetic guarded HTTP responses; QuickChart file storage remains an in-memory fixture. Real provider requests, DNS/TLS, MinIO/S3, ClamAV, hosted organization Review, AI phrasing and independent human confirmation remain unverified.

Vero and Vonage actions

Use an exact saved veroApi {authToken} or vonageApi {apiKey,apiSecret} workspace account. Named fields, Advanced JSON and Format JSON are local; Save uses zero HTTP and stays unverified, and Test is unsupported. No OAuth, new acknowledgment field or generic credential fallback is provided. The operation is selected explicitly; source descriptor defaults describe original intent and do not create missing required input at run time.

Vero: eight confirmed actions

The original resource defaults to user, its user operation to create, and its event operation to track. The retained flattened metadata also records operation:"track" from the event descriptor; it is not a replacement for selecting an operation. Optional selector fields resource and operation, when supplied, must match the selected action.

All paths are relative to fixed https://api.getvero.com/api/v2. Vero uses JSON and query auth_token, an explicit migration from the original ambiguous simultaneous body/form helper, with no legacy fallback. Tags are arrays instead of original stringified arrays. Events use data/extras objects and omit redundant top-level email, rather than copying original JSON strings. Vero may normalize event-name case, spaces and underscores and deduplicate equivalent events within five minutes; no timestamp bypass or exactly-once guarantee is supplied.

id, newId, tags, email, eventName and JSON text originally default to empty strings; required identity/event fields still need valid values. IDs and event names retain whitespace and have a local 254-character limit; event email must be valid. tags may be empty and splits exactly on commas: "new, returning,," retains the leading space and empty entries. No trimming or deduplication is performed.

For create and track, jsonParameters defaults to false. Json Parameters is a switch; Data Attributes Ui is a JSON textarea containing {"dataAttributesValues":[{"key":"plan","value":""}]}. These are string key/value rows, not nested rendered row controls. An absent or empty row collection omits data. For create, Additional Fields is a JSON textarea with optional email; its empty-string default omits email as in the original. For track, Extra Attributes Ui similarly accepts {"extraAttributesValues":[{"key":"source","value":"checkout"}]}.

With JSON mode true, Data Attributes Json and Extra Attributes Json are text inputs whose contents must be JSON objects when present. Leave an unused field absent; a selected empty string is invalid JSON. Nested null, false, empty strings, arrays and exact numeric lexemes retain their values. Large JSON numbers are serialized without passing through a JavaScript Number; put those numbers in JSON text, not numeric expressions that have already lost precision. Malformed JSON, top-level nonobjects, duplicate decoded keys, unsafe prototype keys, ambiguous row duplicates and unsupported fields refuse before HTTP instead of silently omitting or overwriting data. Supplying both UI and JSON sources is refused. New Campaigns, channels/preferences, batch item loops, raw output and continueOnFail parity are not supplied.

Vonage: one SMS send

send uses required From (from) and To (to, E.164 digits, 7–15 digits, no plus). Type defaults to text when omitted; choose text or unicode with Message. Unicode and GSM encoding affect segmentation and fees. binary instead requires literal even-length hex Body and Udh, with a local combined 140-byte limit; do not supply Message or a workspace file. The original hidden Type options also included WAP Push, vCard and vCal. Those choices, their title/url/validity payloads, and unused binaryPropertyName (original data) are refused before HTTP, never silently sent as text.

The fixed POST is https://rest.nexmo.com/sms/json, form encoded with HTTP Basic authentication. This explicitly replaces original form credentials; no legacy fallback, Messages or Verify API substitution is attempted. The original resource/operation defaults are sms/send; From, To and Message defaulted to empty strings and still require meaningful values for their mode.

Additional Fields is one JSON textarea with all seven original option descriptors:

The six executable options preserve false, class zero and empty client-ref. Invalid types or selected empty callback/protocol/class/TTL are not normalized into omission. For example, {"client-ref":"","message-class":0,"protocol-id":"0","status-report-req":false,"ttl":1} keeps those values and sends TTL as 60000 milliseconds. Source defaults are not automatically injected into an absent Additional Fields object.

A callback asks Vonage to send outbound delivery receipts to the exact independently confirmed address. Syntax checks do not verify ownership or control vendor DNS, sockets, redirects, endpoint behavior or delivery. No receiver, polling, receipt proof or delivery cancellation is provided. Existing vendor account callbacks may still operate. SMS fees, trial restrictions, sender/destination permissions and regulatory eligibility remain unverified.

Confirmation, results and retained data

Independently confirm every action, including callback material. Current principal, selected workspace account/revision, direct workspace AND current organization membership, execute permission, live state, workflow/run identity and cancellation are rechecked after awaits; the real transport repeats checks after DNS in onRequest before HTTP. No atomic guarantee spans the database and socket. Organization-stored credentials are unsupported.

One attempt occurs even with outer maxTries3; there are no redirects, retries, polls, readbacks, extra auth probes or returned-link fetches. Vero projects documented status/message; Vonage preserves bounded message/digit/decimal strings without numeric coercion. HTTP 200 plus all Vonage status strings "0" means API acceptance only, not delivery. Nonzero or mixed statuses and all post-invocation failures leave effects uncertain. Every receipt retains outcomeVerified:false; investigate independently before considering another attempt.

Full bounded upstream responses, including discarded fields, enter memory and are scanned before selected projection. Ordinary inputs, selected output, configuration, run history and downstream retention remain. Native limits include a 10-second budget beginning before leaf preparation, 32 KiB request and 256 KiB decoded JSON bounds, depth 16/10000 nodes/8192 reads; outer database/confirmation latency is separate. These are local policies, not vendor, wire/decompression, allocation, OS or socket guarantees.

Local PostgreSQL authority tests cover two accounts, nine operations and current organization/ workspace revocation with synthetic HTTP and confirmation. They establish neither real API acceptance nor delivery, DNS/TLS, independent human consent, hosted Review, natural-language AI interpretation through persistence, production-browser behavior or full n8n parity.

UpLead and OneSimpleAPI: existing accounts and confirmed utilities

UpLead enriches company and person records. OneSimpleAPI provides ten utility definitions: nine support requests and Instagram is explicitly unavailable. These are selected native integrations, not full n8n node, item-loop, credential-import, error-mode or live-provider parity. The original ordinary and Tool resource/operation aliases identify these definitions; they do not import saved credentials or certify all original options.

Choose a manually acquired provider account, then save and select the exact account for the workflow workspace. UpLead uses upleadApi with only apiKey, sent as raw Authorization, never an added Bearer prefix. OneSimpleAPI uses oneSimpleApi with only apiToken, sent as query token with output=json. A provider's individually acquired account is distinct from the application's personal storage scope. Personal accounts can execute through an exact saved-account binding owned by the current actor, with current workspace and organization permissions; real-provider and Review behavior remains unverified. Organization-stored credentials are unsupported. An organization workspace still requires current workspace and organization authority.

Named password fields, Advanced: edit JSON and Format JSON validate locally. Save sends zero provider requests, leaves authentication unverified and grants no execution authority. Test is unavailable; there is no OAuth flow, imported credential-name fallback, extra acknowledgment checkbox or billing control. Never paste account secrets into operation fields or URLs.

Independently confirm every invocation, including metered GET lookups. Review the exact operation, selected account and complete input. Current actor, workspace/organization membership, execute permission, account revision, live preference, run/workflow/version and cancellation are checked again after awaited work and at the guarded transport boundary after provider DNS. These checks do not create an atomic database-to-socket guarantee. Each invocation makes at most one attempt even when workflow maxTries=3 requests retries. A failure after invocation leaves an uncertain outcome; inspect it before manually repeating. outcomeVerified:false means there is no independent verification of the effect, not a retry instruction or a success certificate.

Inputs and selected results can remain in ordinary run history and downstream steps. Full bounded responses, including later-discarded fields, are received and scanned for known secret material before projection; this is not universal secret detection. JSON numeric lexemes are preserved as decimal strings, avoiding JavaScript rounding. Local 10-second/256-KiB and parser/read limits do not guarantee vendor, wire, allocation or operating-system bounds. Account validity, quotas, costs, current provider contracts, hosted Review flows and human confirmation remain unverified.

UpLead input and result choices

company.enrich uses uplead_company_enrich: provide domain or a nonblank company. person.enrich uses uplead_person_enrich: provide email, or nonblank firstname, lastname and domain. Field names keep the original casing. Original resource defaults are company and enrich; the app selects an explicit operation. All identifier controls start as empty text. Original descriptors did not mark these fields required, but the native validator requires one complete lookup alternative. Whitespace is preserved, not silently trimmed into a different lookup; domain/email syntax and bounded strings are validated. An empty optional value stays meaningful, while a wholly empty lookup is rejected before HTTP.

The fixed GET destinations are https://api.uplead.com/v2/company-search and /v2/person-search; identifiers are shared with UpLead and may consume credits. Selected company fields include identity/domain, address/contact data, industry/classification, revenue/employees and social URL metadata; selected person fields include identity, job/contact data, location, employer/domain and social URL metadata. Available-credit data and unknown fields are omitted after scanning the full bounded response. Output is data plus outcomeVerified:false; data may be an object, array or null. A null/no-match result does not prove nonexistence, and optional field presence is not guaranteed. Returned URL fields are never fetched.

OneSimpleAPI operation choices

The original resource defaults are website/pdf, socialProfile/instagramProfile, information/exchangeRate and utility/validateEmail; the app uses explicit definitions. Native credential auth remains query token plus output=json, separate from the input field named output.

Options is one JSON object textarea, not separate nested widgets. A selected boolean stays boolean, including false; omitted options remain omitted. PDF's original selected-option defaults are page empty/force false; screenshot's are screen empty/force false/fullpage false; SEO's headers false. These describe original option selection, not insertion of every option into an empty object. Unknown keys and malformed/type-mismatched values are rejected. Original output was shown only for download:true; the native descriptor exposes it with default data, while binary mode still refuses. It must be an identifier-shaped string and does not rename the native data result envelope.

Text inputs support expressions that retain the resolved type; defaults do not silently stringify an object or number. The download switch is a literal boolean. Operation controls use the original labels and actual native field names; the Help controls and disclosures are translated, but this is not a claim that every raw operation label is localized.

URL disclosure and unavailable modes

PDF, screenshot, SEO, image metadata and URL expansion ask OneSimpleAPI to retrieve the exact approved public HTTPS link. The local syntax policy excludes credentials, fragments, IP literals, private/local names, non-HTTPS and nonstandard ports. Confirmation approves the exact input and current account/run authority before transport; the application does not probe the requested URL, establish target ownership, or control the vendor's DNS, sockets, redirects or remote rendering. No remote-safety guarantee follows from local URL validation. Returned links are metadata and are never followed or downloaded here.

PDF, screenshot and QR require download:false. download:true refuses before HTTP: there is no link substituted as a successful binary download and no workspace file. Instagram also refuses before HTTP because its current contract is unknown; a cached documentation 404 does not prove removal. Original QR Small/Medium/Large and PNG/SVG casing and image raw=true are retained but current vendor acceptance/semantics remain unknown. Screenshot fullpage explicitly maps to documented full=yes/no. Only the listed original options are accepted; current vendor features such as PDF background/HTML, QR colors or other output formats are not added implicitly.

Original descriptors and current native controls

The tables preserve original defaults, choices and display conditions. Original visibility is migration evidence; the actual native controls listed separately do not recreate every conditional n8n widget. No original JavaScript is executed.

uplead company.enrich

Independent synthetic selected-output example; this is not an observed provider response:

{
  "data": {
    "id": "company-A",
    "company_name": "Alpha",
    "domain": "alpha.example"
  },
  "outcomeVerified": false
}

uplead person.enrich

Independent synthetic selected-output example; this is not an observed provider response:

{
  "data": {
    "id": "person-A",
    "email": "a@example.com",
    "email_status": "valid"
  },
  "outcomeVerified": false
}

onesimpleapi pdf

Independent synthetic selected-output example; this is not an observed provider response:

{
  "data": {
    "page": "A4",
    "url": "https://example.com/generated-A.pdf"
  },
  "outcomeVerified": false
}

onesimpleapi screenshot

Independent synthetic selected-output example; this is not an observed provider response:

{
  "data": {
    "url": "https://example.com/generated-A.png",
    "width": "414",
    "height": "828"
  },
  "outcomeVerified": false
}

onesimpleapi seo

Independent synthetic selected-output example; this is not an observed provider response:

{
  "data": {
    "general": {
      "title": "Alpha"
    },
    "headers": {
      "content-type": "text/html"
    }
  },
  "outcomeVerified": false
}

onesimpleapi instagramProfile

Independent synthetic selected-output example; this is not an observed provider response:

null

onesimpleapi spotifyArtistProfile

Independent synthetic selected-output example; this is not an observed provider response:

{
  "data": {
    "name": "Alpha",
    "followers": "10",
    "type": "artist"
  },
  "outcomeVerified": false
}

onesimpleapi exchangeRate

Independent synthetic selected-output example; this is not an observed provider response:

{
  "data": {
    "from_currency": "USD",
    "from_value": "10.25",
    "to_currency": "EUR",
    "to_value": "9.5"
  },
  "outcomeVerified": false
}

onesimpleapi imageMetadata

Independent synthetic selected-output example; this is not an observed provider response:

{
  "data": {
    "make": "Camera A",
    "latitude": "1.25"
  },
  "outcomeVerified": false
}

onesimpleapi expandURL

Independent synthetic selected-output example; this is not an observed provider response:

{
  "data": {
    "url": "https://example.com/expanded-a",
    "hops": "1"
  },
  "outcomeVerified": false
}

onesimpleapi qrCode

Independent synthetic selected-output example; this is not an observed provider response:

{
  "data": {
    "url": "https://example.com/qr-a.png"
  },
  "outcomeVerified": false
}

onesimpleapi validateEmail

Independent synthetic selected-output example; this is not an observed provider response:

{
  "data": {
    "deliverability": "GOOD",
    "is_format_valid": true,
    "safe_to_register_as_user": true
  },
  "outcomeVerified": false
}

LoneScale and PostBin: fields, migrations and outputs

LoneScale uses native lonescaleApi (lowercase s) with exactly apiKey. Explicitly resave original loneScaleApi; there is no generic alias or fallback. Requests send x-api-key only to public-api.lonescale.com. The original overlapping Authorization/X-API-KEY intent is replaced explicitly: current endpoint security schemas specify x-api-key, although overview prose mentions Authorization. Save, reopen, named fields, Advanced JSON and Format send zero provider HTTP and leave the account unverified. Test, OAuth, credential discovery and key generation are unavailable. Select the exact saved account for the workflow workspace.

PostBin has authentication none and credentials:[]. Do not create or select a saved account, key, credential form, Test connection or OAuth flow. Its public bin/request IDs establish neither ownership nor privacy; use only bins and data you are authorized to access.

createList takes required name and Type (COMPANY by default, or PEOPLE). It sends name and entity. addItem defaults to PEOPLE and requires a list UUID. Original people mode requires first_name and last_name, plus optional peopleAdditionalFields: full_name, email, company_name, current_position, domain, linkedin_url, location, contact_id. Original company mode uses company_name, separate native name, and optional companyAdditionalFields: linkedin_url, domain, location, contact_id.

Alternatively supply fields JSON alone, without original top-level identity fields or either additional-fields collection. Supported keys are first_name, last_name, full_name, linkedin_url, company_name, current_position, domain, location, name, email, contact_id. PEOPLE requires two nonempty documented identity fields from linkedin_url, name, location, first_name, last_name, full_name, company_name, current_position; email/domain/contact_id alone do not satisfy this rule. COMPANY requires domain OR separate name and location. company_name never silently becomes name. Entity-incompatible fields and mixed input modes refuse before HTTP. Strings preserve case, spaces and numeric-looking values such as "0007"; empty optional strings follow original truthy-string omission. Email and LinkedIn URL syntax is validated; this does not prove identity.

listOptions replaces the automatic list loader with one explicit confirmed GET: choose Type and zero-based page (default 0, maximum 1000000). The documented page contains ten records; selected options filter entity and preserve hasNextPage, total and incompleteness. Supply the next page explicitly; there is no automatic pagination or Save/Test request. Workflow options, workflow-hook create/check/delete and LoneScale triggers remain unavailable because current route/authentication/output, callback authenticity and durable ownership contracts are unresolved. GET /webhooks is a different contract and is never substituted.

PostBin supports createBin, getBin, deleteBin, getRequest, removeFirstRequest and sendRequest. Bin creation has no required content. Get/delete require binId; get request also requires requestId; consume-first requires binId; send requires binId and defaults binContent to an empty string. Use a finite safe opaque ID (ASCII letters, digits, _ or -, at most128 characters, beginning with a letter or digit), or an exact supported original numeric bin URL/label. This explicit migration accepts documented opaque IDs as well as the original numeric form. Bin IDs api/b and request ID shift are refused where they would select another route. A read alias never consumes the first request.

removeFirstRequest uses GET /api/bin/{binId}/req/shift but removes data: treat it as an external write and confirm it independently. deleteBin uses DELETE; HTTP200 does not prove that the bin previously existed. sendRequest POSTs native JSON {"content":string} to the exact bin: this is an explicit serializer migration, not proof of the original generic serializer. Empty text is intentional. Stored false, null, zero, empty strings/arrays/objects remain distinct; selected JSON number lexemes become exact strings without rounding. Bin results expose selected ID, numeric timestamps/ISO forms and constructed request/view URLs. Request results expose selected method/path/headers/query/body/IP/timestamps; send returns a request ID. Returned links are never followed and no readback proves delivery.

Request Options is one JSON-object textarea, not separate native switches. All four original options have an explicit disposition:

The actual HOST item plan validates every item before the first HTTP request, accepts at most100 items with the same pacing policy, and executes sequentially. Aggregate interbatch waiting is at most30000 ms. It confirms and rechecks current authority for each item, pairs each result with its input and stops after failure while retaining prior receipts in failed output. It does not automatically resume or provide durable continuation. The custom timeout starts with each native invocation; outer pacing, confirmation and database latency are separate. Local limits do not prove wire/allocation or operating-system confinement.

Every action, including reads and list options, needs independent confirmation of its exact input. Each action has one attempt even with imported maxTries:3. Successful and handled-failure receipts keep outcomeVerified:false; a failed write after invocation may already have changed the provider. Investigate uncertainty before repeating. Error routing does not turn failure into success, retry or compensation. Inputs, selected outputs, run history and downstream data use ordinary retention. Stored PostBin headers, body and IP can reveal personal data.

Current actor, workspace AND organization membership, execute permission, selected account revision where applicable, live run and cancellation are checked again around asynchronous work. This does not provide an atomic database/socket guarantee. Workspace accounts and explicitly selected personal native accounts resolve by exact credential ID and credential family. A personal account must belong to the authenticated run actor in the workflow workspace, with current workspace AND organization permission; its display name grants no authority and cannot select another account. Organization-stored credentials are unsupported. Missing or changed persisted run actors refuse before dispatch. A personally acquired provider key is distinct from its application storage scope. Local and synthetic tests do not prove real provider acceptance, organization persistence, independent human consent or Review behavior. No full n8n parity or durable continuation is claimed.

LoneScale actual native controls

The current inspector exposes these controls together. Original conditional visibility is retained below as migration context; it is not a claim of identical native visibility.

createList — Create list

Synthetic literal input, not a live-provider result:

{
  "name": "Companies A",
  "type": "COMPANY"
}

Selected synthetic output expectations:

{
  "expectedBodyText": "{\"name\":\"Companies A\",\"entity\":\"COMPANY\"}",
  "expectedKnownOutput": {
    "id": "11111111-1111-4111-8111-111111111111",
    "name": "Companies A",
    "entity": "COMPANY",
    "organizationId": "org-A"
  }
}

addItem — Add item

Synthetic literal input, not a live-provider result:

{
  "type": "PEOPLE",
  "list": "11111111-1111-4111-8111-111111111111",
  "first_name": "Ada",
  "last_name": "Lovelace",
  "peopleAdditionalFields": {
    "email": "ada@example.test",
    "company_name": "Example",
    "contact_id": "0007"
  }
}

Selected synthetic output expectations:

{
  "expectedBodyText": "{\"first_name\":\"Ada\",\"last_name\":\"Lovelace\",\"company_name\":\"Example\",\"email\":\"ada@example.test\",\"contact_id\":\"0007\"}",
  "expectedKnownOutput": {
    "id": "aaaaaaaa-aaaa-4aaa-8aaa-aaaaaaaaaaaa",
    "first_name": "Ada",
    "last_name": "Lovelace",
    "email": "ada@example.test",
    "contact_id": "0007"
  }
}

listOptions — Get list-options page

Synthetic literal input, not a live-provider result:

{
  "type": "PEOPLE",
  "page": 0
}

Selected synthetic output expectations:

{
  "expectedOptions": [
    {
      "name": "P",
      "value": "11111111-1111-4111-8111-111111111111"
    }
  ]
}

LoneScale original fields and option choices

Defaults and visibility below describe inert shipped metadata. Finite action/resource constants map to the selected native operation; unsupported settings are refused as described above. Original loaders are replaced only by the explicit list-options operation.

PostBin actual native controls

The current inspector exposes these controls together. Original conditional visibility is retained below as migration context; it is not a claim of identical native visibility.

createBin — Create bin

Synthetic literal input, not a live-provider result:

{}

Selected synthetic output expectations:

{
  "expectedBody": null,
  "expectedNativeOutput": {
    "binId": "1792539000000-1792539000001",
    "nowTimestamp": 0,
    "nowIso": "1970-01-01T00:00:00.000Z",
    "expiresTimestamp": 1800000,
    "expiresIso": "1970-01-01T00:30:00.000Z",
    "requestUrl": "https://www.postb.in/1792539000000-1792539000001",
    "viewUrl": "https://www.postb.in/b/1792539000000-1792539000001"
  }
}

getBin — Get bin

Synthetic literal input, not a live-provider result:

{
  "binId": "1792539000000-1792539000001"
}

Selected synthetic output expectations:

{
  "expectedBody": null,
  "expectedSelectedTimestamps": {
    "nowTimestamp": 0,
    "nowIso": "1970-01-01T00:00:00.000Z",
    "expiresTimestamp": 1800000,
    "expiresIso": "1970-01-01T00:30:00.000Z"
  }
}

deleteBin — Delete bin

Synthetic literal input, not a live-provider result:

{
  "binId": "1792539000000-1792539000001"
}

Selected synthetic output expectations:

{
  "expectedBody": null,
  "expectedKnownOutput": {
    "msg": "Bin deleted"
  },
  "expectedExistenceClaim": false
}

getRequest — Get request

Synthetic literal input, not a live-provider result:

{
  "binId": "1792539000000-1792539000001",
  "requestId": "0007"
}

Selected synthetic output expectations:

{
  "expectedBody": null,
  "expectedStoredBody": {
    "content": "",
    "ok": false,
    "n": 0,
    "tags": []
  }
}

removeFirstRequest — Consume first request

Synthetic literal input, not a live-provider result:

{
  "binId": "1792539000000-1792539000001"
}

Selected synthetic output expectations:

{
  "expectedBody": null,
  "expectedEffect": "external-write",
  "expectedStoredBody": {
    "content": "first"
  }
}

sendRequest — Send request

Synthetic literal input, not a live-provider result:

{
  "binId": "1792539000000-1792539000001",
  "binContent": ""
}

Selected synthetic output expectations:

{
  "expectedBodyText": "{\"content\":\"\"}",
  "expectedKnownOutput": {
    "requestId": "0007"
  }
}

PostBin original fields and option choices

Defaults and visibility below describe inert shipped metadata. Finite action/resource constants map to the selected native operation; unsupported settings are refused as described above.

Stackby, Dropcontact and Oura accounts

Save an existing Stackby key as stackbyApi or a Dropcontact key as dropcontactApi, with exactly {"apiKey":"YOUR_EXISTING_KEY"}. Stackby uses the API Beta profile at https://stackby.com/api/betav1/ with api-key; other Stackby profiles are separate contracts and are not tried as fallbacks. Dropcontact uses https://api.dropcontact.com/v1/enrich/all with X-Access-Token; original .io/batch requests require this explicit migration.

Oura uses the new ouraOAuthAccessToken family with exactly {"authType":"oauth_access_token","accessToken":"YOUR_EXISTING_OAUTH_ACCESS_TOKEN"}. Supply an existing OAuth access token yourself. Retired Personal Access Tokens and original ouraApi credentials are unsupported; resaving opaque bytes does not establish token type, expiry or consent. Profile reads require personal; daily scores require daily. These are required grants, not verified consent or scope metadata. OAuth acquisition, refresh, introspection and callbacks are unavailable. Stackby and Dropcontact retain only their selected ordinary and Tool operation aliases. Oura is a separate manual-token counterpart with no n8n import aliases, including Tool aliases; migrate its inputs and account explicitly.

Use named fields or Advanced JSON; Format validates and formats locally. Save and reopen send zero provider requests and leave authentication unverified. Test connection is unavailable. Select the exact saved account for this workflow: workspace accounts or a personal account owned by the current actor can be used. Current workspace and organization permissions still apply; organization-stored accounts are unsupported. A queued run needs its original persisted actor, and changing an account, actor, workflow version or permission can invalidate an earlier approval.

Confirm the exact input independently for every physical request, including reads, later pages and status checks. Each request has one attempt even with imported maxTries:3. outcomeVerified:false remains explicit after success or a handled failure; a failed write may already have changed the provider. Investigate uncertainty before repeating. Inputs, selected results, ordinary history and downstream steps may retain data. Local and simulated checks do not establish real provider compatibility, token grants or an authenticated human decision.

Stackby rows and paging

Choose row.append, row.delete, row.list or row.read. Supply a Stack ID and table name. Read also takes a record ID; delete takes an array of one to ten row IDs. Append accepts one to ten records, each with a field object, or Original items with each item's Stack ID, table, comma-separated columns and item JSON. Original-item migration preserves destination grouping and input indices, splits physical writes into groups of at most ten rows, and retains every batch receipt. A partial failure does not erase earlier receipts or authorize automatic replay. Read, list and append require outputFields, an array of one to fifty field names; missing vendor values remain missing, separately from false, null and empty values.

List takes an explicit Page or Complete mode. Its original returnAll:true and limit:1000 defaults express intent; they do not establish completeness. Native controls include offset (default 0, maximum 1000000), page size (default 100, range 1..100), limit (1..1000) and view (default empty text). Page mode returns an explicitly incomplete page. Complete migration persists its plan and waits between windows; every later physical page needs a fresh independent confirmation. It stops on an observed empty page or a local bound, without inventing a total or guaranteeing all provider rows. Process-local pacing cannot control other clients. API Beta compatibility remains unverified with a real account.

Dropcontact enrichment and status

Choose enrich to submit a contact or fetchRequest to read one exact requestId. Enrichment has an Email field (default empty text), Additional Fields JSON (default {}) and Options JSON (default {}). Additional fields are num_siren, siret, company, country, first_name, full_name, last_name, linkedin, phone and website. Options accept waitTime in seconds (1..86400, effective default 45), siren and language (en or fr). Original defaults are Simplify false, French company enrichment false and language English. Preserve explicit false and omitted options separately.

Enable Simplify (submit job only) for a single submission. Its request ID does not establish completed enrichment. With the original Simplify false default, the workflow persists a configurable timer (45 seconds by default), then opens a separate Human Task before one confirmed status GET. A pending result stays pending; there is no hidden polling. Completion migration needs the durable runtime and Human Tasks and refuses before sending inside loops, parallel blocks, subworkflows and single-step execution. Submit-only and explicit status reads are separate available choices.

A custom public HTTPS callback URL asks Dropcontact to deliver contact data to that exact address. Existing account callbacks may also receive it. Review this sharing before confirmation; the application neither probes the URL nor establishes receiver ownership, vendor DNS, redirects or delivery. Contact requests may consume credits. The ten-second physical-request budget excludes the outer timer, human decision and database waits.

Oura profile and daily scores

Choose profile.get, summary.getActivity, summary.getReadiness or summary.getSleep. Profile has no input fields and excludes email from its selected result. Sleep returns a daily sleep score, rather than sleep periods. Profile and summary numeric JSON values are returned as exact decimal text; omitted optional fields remain omitted and null remains null.

Every summary requires explicit startDate and endDate calendar dates in YYYY-MM-DD, with at most 31 inclusive days. This replaces the original timezone-dependent datetime and omitted-date defaults. Limit is 1..10 (default 5), applied locally and never sent to Oura. Optional nextToken selects one explicit provider page; there is no Return All or automatic pagination. receivedCount, returnedCount, omittedCount and locallyTruncated describe only that page. upstreamHasNextPage and nextToken describe upstream continuation: advancing the token cannot recover rows omitted locally from the preceding page. Raising the limit and rerunning requires another confirmation and is not a completeness guarantee.

Oura receives personal and health-data requests. The full bounded upstream response, including fields later discarded, enters memory before projection; selected outputs and ordinary history/downstream retention remain. No live token, identity, subscription, consent, scope or health-data accuracy verification is implied. Flow remains research-only because authority for its former API could not be established; it is not an available integration.

Native field reference: Stackby, Dropcontact and Oura

The following fields belong to the selected native operation. Fixed resource/operation values describe that operation; they do not select a different action. Typed fields and JSON collections preserve their declared types; a complete expression must return the matching type. Oura limit is a numeric control. Account selection and independent confirmation are separate from these inputs.

stackby row.append

stackby row.delete

stackby row.list

stackby row.read

dropcontact enrich

dropcontact fetchRequest

ouraRead profile.get

ouraRead summary.getActivity

ouraRead summary.getReadiness

ouraRead summary.getSleep

Zammad, Wekan, Quickbase and Salesmate accounts

Select the exact saved account and matching credential family in the workflow. Workspace accounts and personal accounts owned by the current actor are eligible; organization-stored accounts are unsupported. Current workspace and organization execution permissions still apply. Saving does not authorize a request or verify token identity, provider rights or consent. Named fields, Advanced JSON and Format validate the strict local shape; unknown, duplicate or invalid fields are refused without silently repairing the input. Save and reopen send no provider request and leave authentication unverified. Test connection, automatic login and OAuth acquisition or refresh are unavailable for these profiles.

Zammad: explicit Basic or Token mode

Use zammadNativeBasicApi for Basic operations:

{
  "mode": "basicAuth",
  "baseUrl": "https://support.example.com",
  "username": "YOUR_EXISTING_USERNAME",
  "password": "YOUR_EXISTING_PASSWORD"
}

Use zammadNativeTokenApi for Token operations:

{ "mode": "tokenAuth", "baseUrl": "https://support.example.com", "accessToken": "YOUR_EXISTING_TOKEN" }

Choose the corresponding basicAuth.<action> or tokenAuth.<action> operation: 25 logical actions become 50 mode-qualified choices. Use the public HTTPS installation root or supported path prefix. Token mode uses Zammad's Token token= authorization. OAuth Bearer and original zammadBasicAuthApi or zammadTokenAuthApi credentials require explicit migration; no family or server fallback is tried. No n8n import aliases are declared.

Use names where a field asks for names: organization and decimal organizationId are mutually exclusive. Create needs group/organization name, user firstname and lastname, or ticket title, group, customer and article. Nested fields JSON must follow the selected resource shape; empty updates and unsafe or duplicate custom field names are refused. Supply the selected installation's custom-field names manually.

Lists, user/field options and article lists request one explicit page: page defaults to 1, perPage to 50, maximum 100. Every later page or option read needs a separate confirmation; nextPageCandidate is unverified and complete:false does not promise an exhaustive result. Migrate Return All or larger imported limits manually. Ticket creation returns its POST response without an article readback; confirm article.list separately. Article upload and attachment download use actual owned workspace files, at most 256 KiB, with current ownership, scope, ready scan state and integrity checks. Download requires ticket/article/attachment IDs, file name and MIME type. A URL is not a file and is never followed implicitly.

Wekan: existing Bearer token and selected user

Save wekanExistingBearerApi with exactly:

{
  "accountOrigin": "https://boards.example.com",
  "token": "YOUR_EXISTING_BEARER_TOKEN",
  "userId": "YOUR_EXISTING_USER_ID"
}

Use the public HTTPS installation origin or supported path prefix and supply the original opaque userId yourself. Declaring a user ID does not prove token ownership. Original wekanApi username/password credentials need explicit migration. The selected static contract targets Wekan v12.57; installed-server capability and permissions remain unverified.

There are 35 choices: 24 ordinary actions, three explicitly named runtimeChecklist actions and eight independent options reads. Only the 24 ordinary actions declare ordinary Wekan aliases; Tool aliases, triggers and attachment operations are unavailable. Each options read needs separate confirmation, and listing users may require server administrator rights.

Board owner and card/comment create author must match the selected user. Board role overrides and later card-author changes are refused. Use opaque IDs, not routes or encoded separators. For card.getAll, provide the list or swimlane ID selected by fromObject. Collections return one response, locally limited to 1–200 records, default 100; returnAll still means this bounded response, completeness is unknown and nextPage is null. There is no cursor or hidden pagination. Explicit false, zero, empty text and omission remain distinct locally. Card dates support null/empty clearing and members/assignees support empty arrays, but the server may ignore falsy updates, filter inactive members or truncate an updated card title to 1,000 characters.

Quickbase: US realm and explicit field IDs

Save quickbaseApi with exactly:

{ "authType": "user_token", "hostname": "your-realm.quickbase.com", "userToken": "YOUR_EXISTING_USER_TOKEN" }

hostname is the exact US realm hostname, without scheme or path. Requests use fixed https://api.quickbase.com/v1; EU/custom origins are unavailable. Resave original QuickBaseApi credentials explicitly. Ten ordinary actions retain ordinary and Tool aliases; getTableFields and getUniqueTableFields are two separate first-class discovery actions without aliases.

Supply table and field IDs manually. Runtime requirements still apply even when the shared form marks a field optional: a table request needs tableId, and record writes need the selected fields/items. Record writes require options.useFieldIDs:true; Update also requires simple:false and explicit Record ID field 3 matching updateKey. Hidden field discovery followed by a write and multiple physical request units are unavailable. Discover fields in a separate approved action, then configure IDs. Create, Update and Upsert share the vendor insert/update endpoint and may insert or update.

A batch has 1–10 original items, sharing item 0 table/options and per-item columns; in explicit-ID mode every item key is sent, regardless of columns. Keep itemParameters and itemBinary cardinality/order aligned. Missing or ambiguous labels/IDs and collisions refuse rather than inventing values. Each physical request, including reads and discovery, needs its own signed exact-request Human Task approval by the original actor. Review method, URL, realm, query, complete body and input indexes in the preview, at most 48 KiB. The initial path permits one physical request. Generic Run or single-step acknowledgment cannot replace this approval.

Reads return one bounded page, up to 100 records; Return All is still one page. Advance skip manually after reviewing incomplete/stalled/nextCandidate metadata; there is no snapshot or exhaustive-result guarantee. Raw envelopes remain available and JSON numeric lexemes stay exact strings; HTTP 207 line errors can mean a partial result. Downloads decode at most 1 MiB of actual bytes into owned workspace files with scan, size and SHA256 checks; returned URLs are not fetched as files.

Salesmate: existing v4 tenant token and flat CRUD fields

Save salesmateV4AccessToken with exactly:

{
  "profile": "salesmate-v4-access-token",
  "tenantOrigin": "https://your-tenant.salesmate.io",
  "accessToken": "YOUR_EXISTING_V4_ACCESS_TOKEN"
}

Use the exact lowercase HTTPS tenant origin without trailing slash, path, port, credentials, query or fragment. Central/reserved hosts, legacy salesmateApi/sessionToken and v1/v2 endpoints require explicit migration; no fallback is tried. The durable runtime with persisted authority is required; memory-only execution cannot authorize these requests.

Twelve actions create, get, update or delete one activity, company or deal. No n8n import aliases are declared. GetAll/search, custom fields, activity dueDate, bulk IDs, hidden loaders and the old Get projection are unavailable. Get requires rawData:true; create/update default to rawData:false. Get/update/delete IDs are positive decimal text, up to 20 digits. owner, primaryContact and primaryCompany use the declared numeric controls. A provider owner ID grants no app authority.

Updates require all mandatory creation fields: activity title, owner, type; company name, owner; deal title, owner, primaryContact, current pipeline/stage names, status (Open, Close or Lost) and uppercase three-letter currency. Supply pipeline/stage names manually. Migrate the original Billing State/Country collision into separate billingState and billingCountry; overwritten values cannot be reconstructed. Explicit false and zero remain supported where declared.

Author dealValue as decimal text, including expressions returning text: at most 18 total digits and nine fractional digits, optional minus sign, no plus or exponent. Preserve digits, scale, zero and negative zero. Empty text is invalid; omission remains omitted. Migrate imported numeric workflow values explicitly; already-rounded IEEE754 digits cannot be recovered. For example, use the string "123456789.123456789" without converting through Number.

Every action, including Get, requires its own durable exact-request Human Task decision by the original persisted actor. Generic Run approval and provider owner IDs are insufficient. Input preview is bounded to 24 KiB, the signed payload to 32 KiB, and approval expires after 15 minutes. Account, configuration or authority changes can invalidate it sooner. CRM writes may invite/notify people, trigger account automations and consume quota or credits. Delete is destructive; the activity soft-delete request choice does not guarantee recovery. Returned provider JSON retains numeric lexemes as strings; links remain data, without file transfer or implicit fetches.

Confirm each request and inspect uncertain effects

All four profiles require independent confirmation and current account, original actor, workspace and organization authority, workflow/configuration revision and live-run/cancellation checks. Each invocation has one attempt even when workflow retries are enabled. No automatic retry, polling, readback or compensation verifies a change. outcomeVerified:false leaves effects unverified after HTTP success; inspect uncertain changes before intentionally repeating them. Inputs, full bounded provider data, selected results and ordinary history/downstream data may persist. Saving an account or accepting a confirmation does not prove real-provider compatibility, permissions, scanning or delivery. These selected counterparts do not establish complete n8n action, field, authentication, import or runtime parity.

Native field reference: Zammad, Wekan, Quickbase and Salesmate

The 109 concrete operations below expose 417 top-level controls. Required/default describes the actual local form; runtime rules in the provider sections still apply. Text expressions are supported only where marked. Nested JSON accepts only the selected schema, not arbitrary replacement fields. No control runs hidden discovery.

zammad basicAuth.group.create

Account family: zammadNativeBasicApi.

zammad basicAuth.group.delete

Account family: zammadNativeBasicApi.

zammad basicAuth.group.get

Account family: zammadNativeBasicApi.

zammad basicAuth.group.getAll

Account family: zammadNativeBasicApi.

zammad basicAuth.group.update

Account family: zammadNativeBasicApi.

zammad basicAuth.organization.create

Account family: zammadNativeBasicApi.

zammad basicAuth.organization.delete

Account family: zammadNativeBasicApi.

zammad basicAuth.organization.get

Account family: zammadNativeBasicApi.

zammad basicAuth.organization.getAll

Account family: zammadNativeBasicApi.

zammad basicAuth.organization.update

Account family: zammadNativeBasicApi.

zammad basicAuth.ticket.create

Account family: zammadNativeBasicApi.

zammad basicAuth.ticket.delete

Account family: zammadNativeBasicApi.

zammad basicAuth.ticket.get

Account family: zammadNativeBasicApi.

zammad basicAuth.ticket.getAll

Account family: zammadNativeBasicApi.

zammad basicAuth.user.create

Account family: zammadNativeBasicApi.

zammad basicAuth.user.delete

Account family: zammadNativeBasicApi.

zammad basicAuth.user.get

Account family: zammadNativeBasicApi.

zammad basicAuth.user.getAll

Account family: zammadNativeBasicApi.

zammad basicAuth.user.getSelf

Account family: zammadNativeBasicApi.

No operation input fields.

zammad basicAuth.user.update

Account family: zammadNativeBasicApi.

zammad basicAuth.user.options

Account family: zammadNativeBasicApi.

zammad basicAuth.field.options

Account family: zammadNativeBasicApi.

zammad basicAuth.article.list

Account family: zammadNativeBasicApi.

zammad basicAuth.article.upload

Account family: zammadNativeBasicApi.

zammad basicAuth.attachment.download

Account family: zammadNativeBasicApi.

zammad tokenAuth.group.create

Account family: zammadNativeTokenApi.

zammad tokenAuth.group.delete

Account family: zammadNativeTokenApi.

zammad tokenAuth.group.get

Account family: zammadNativeTokenApi.

zammad tokenAuth.group.getAll

Account family: zammadNativeTokenApi.

zammad tokenAuth.group.update

Account family: zammadNativeTokenApi.

zammad tokenAuth.organization.create

Account family: zammadNativeTokenApi.

zammad tokenAuth.organization.delete

Account family: zammadNativeTokenApi.

zammad tokenAuth.organization.get

Account family: zammadNativeTokenApi.

zammad tokenAuth.organization.getAll

Account family: zammadNativeTokenApi.

zammad tokenAuth.organization.update

Account family: zammadNativeTokenApi.

zammad tokenAuth.ticket.create

Account family: zammadNativeTokenApi.

zammad tokenAuth.ticket.delete

Account family: zammadNativeTokenApi.

zammad tokenAuth.ticket.get

Account family: zammadNativeTokenApi.

zammad tokenAuth.ticket.getAll

Account family: zammadNativeTokenApi.

zammad tokenAuth.user.create

Account family: zammadNativeTokenApi.

zammad tokenAuth.user.delete

Account family: zammadNativeTokenApi.

zammad tokenAuth.user.get

Account family: zammadNativeTokenApi.

zammad tokenAuth.user.getAll

Account family: zammadNativeTokenApi.

zammad tokenAuth.user.getSelf

Account family: zammadNativeTokenApi.

No operation input fields.

zammad tokenAuth.user.update

Account family: zammadNativeTokenApi.

zammad tokenAuth.user.options

Account family: zammadNativeTokenApi.

zammad tokenAuth.field.options

Account family: zammadNativeTokenApi.

zammad tokenAuth.article.list

Account family: zammadNativeTokenApi.

zammad tokenAuth.article.upload

Account family: zammadNativeTokenApi.

zammad tokenAuth.attachment.download

Account family: zammadNativeTokenApi.

wekan board.create

Account family: wekanExistingBearerApi.

wekan board.delete

Account family: wekanExistingBearerApi.

wekan board.get

Account family: wekanExistingBearerApi.

wekan board.getAll

Account family: wekanExistingBearerApi.

wekan cardComment.create

Account family: wekanExistingBearerApi.

wekan cardComment.delete

Account family: wekanExistingBearerApi.

wekan cardComment.get

Account family: wekanExistingBearerApi.

wekan cardComment.getAll

Account family: wekanExistingBearerApi.

wekan card.create

Account family: wekanExistingBearerApi.

wekan card.delete

Account family: wekanExistingBearerApi.

wekan card.get

Account family: wekanExistingBearerApi.

wekan card.getAll

Account family: wekanExistingBearerApi.

wekan card.update

Account family: wekanExistingBearerApi.

wekan checklist.create

Account family: wekanExistingBearerApi.

wekan checklist.delete

Account family: wekanExistingBearerApi.

wekan checklist.get

Account family: wekanExistingBearerApi.

wekan checklist.getAll

Account family: wekanExistingBearerApi.

wekan checklistItem.delete

Account family: wekanExistingBearerApi.

wekan checklistItem.get

Account family: wekanExistingBearerApi.

wekan checklistItem.update

Account family: wekanExistingBearerApi.

wekan list.create

Account family: wekanExistingBearerApi.

wekan list.delete

Account family: wekanExistingBearerApi.

wekan list.get

Account family: wekanExistingBearerApi.

wekan list.getAll

Account family: wekanExistingBearerApi.

wekan runtimeChecklist.getCheckItem

Account family: wekanExistingBearerApi.

wekan runtimeChecklist.deleteCheckItem

Account family: wekanExistingBearerApi.

wekan runtimeChecklist.updateCheckItem

Account family: wekanExistingBearerApi.

wekan options.users

Account family: wekanExistingBearerApi.

wekan options.boards

Account family: wekanExistingBearerApi.

wekan options.lists

Account family: wekanExistingBearerApi.

wekan options.swimlanes

Account family: wekanExistingBearerApi.

wekan options.cards

Account family: wekanExistingBearerApi.

wekan options.checklists

Account family: wekanExistingBearerApi.

wekan options.checklistItems

Account family: wekanExistingBearerApi.

wekan options.comments

Account family: wekanExistingBearerApi.

quickbase field.getAll

Account family: quickbaseApi.

quickbase file.delete

Account family: quickbaseApi.

quickbase file.download

Account family: quickbaseApi.

quickbase record.create

Account family: quickbaseApi.

quickbase record.upsert

Account family: quickbaseApi.

quickbase record.delete

Account family: quickbaseApi.

quickbase record.getAll

Account family: quickbaseApi.

quickbase record.update

Account family: quickbaseApi.

quickbase report.get

Account family: quickbaseApi.

quickbase report.run

Account family: quickbaseApi.

quickbase getTableFields

Account family: quickbaseApi.

quickbase getUniqueTableFields

Account family: quickbaseApi.

salesmate activity.create

Account family: salesmateV4AccessToken.

salesmate activity.delete

Account family: salesmateV4AccessToken.

salesmate activity.get

Account family: salesmateV4AccessToken.

salesmate activity.update

Account family: salesmateV4AccessToken.

salesmate company.create

Account family: salesmateV4AccessToken.

salesmate company.delete

Account family: salesmateV4AccessToken.

salesmate company.get

Account family: salesmateV4AccessToken.

salesmate company.update

Account family: salesmateV4AccessToken.

salesmate deal.create

Account family: salesmateV4AccessToken.

salesmate deal.delete

Account family: salesmateV4AccessToken.

salesmate deal.get

Account family: salesmateV4AccessToken.

salesmate deal.update

Account family: salesmateV4AccessToken.

MISP accounts and exact requests

Save an existing MISP account as mispApi with exactly:

{ "apiBaseUrl": "https://misp.example.org/security/misp", "apiKey": "YOUR_EXISTING_RAW_API_KEY" }

Use the public HTTPS installation URL, including its exact path prefix, and an existing raw API key. Save and reopen are local and leave authentication unverified. Test connection, OAuth and automatic key acquisition are unavailable. Select the exact workspace account or a personal account owned by the run actor. Current workspace and organization execution permissions apply; organization-stored accounts are unsupported.

The 44 choices cover attributes, events, event tags, feeds, galaxies, notice lists, objects, organisations, tags, users and warning lists. Ordinary MISP and MISP Tool imports identify these covered operations; they do not establish general n8n Tool runtime compatibility. Supply IDs manually as positive decimal text with at most 18 digits. UUIDs and larger IDs require an explicit migration. Additional Fields and Update Fields accept only the declared keys; updates must contain a supported change.

For attribute, event and object search, leave Use Json off and fill Value plus optional Additional Fields. To use Json Output, turn Use Json on and clear Value and Additional Fields. JSON accepts the finite search fields value, category, deleted, searchall, tags, type, published, page, limit and returnFormat:"json"; it is not an arbitrary request body. Explicit false values remain false. GetAll returns one bounded response, with a local limit of 1–1,000 (default 50). Return All does not guarantee completeness or request another page. There are no hidden option lookups, retries or readbacks.

The original user.invited_by, organisation.usercount, organisation.created_by_email and feed-creation additionalFields.json effects are unsupported and refused. Feed update rules use their declared finite JSON shape. Insecure TLS, binary workflows and triggers are unavailable. Original source versions and the selected current API contract differ; installed-instance permissions and provider compatibility remain unverified.

Every read and write pauses for a separate exact-request Human Task assigned to the original run actor. Review the selected account, method, URL, complete body and described effect. Approval expires after 15 minutes and can become invalid sooner if the account, workflow, run, input or permissions change. The preview is limited to 24 KiB and its signed payload to 32 KiB. Another Member or Owner cannot approve on the original actor's behalf. Durable workflow execution and Human Tasks must be enabled.

Publishing an event may queue publication and fan out to other systems. Creating a user may email credentials and create an API key. Deletion can remove provider data. Each confirmed request has one attempt even when workflow retries are enabled. outcomeVerified:false remains explicit: a receipt does not independently verify the change, queued publication or delivery. After an uncertain failure, inspect the provider before intentionally repeating the request. Inputs and selected provider data may remain in run history and downstream steps; credential-like response fields are excluded, but ordinary content can still contain sensitive data.

MISP field reference

The following fields belong to the selected native operations. JSON fields accept only the declared finite keys; branch requirements still apply to search.

misp attribute.create

Account family: mispApi.

misp attribute.delete

Account family: mispApi.

misp attribute.get

Account family: mispApi.

misp attribute.getAll

Account family: mispApi.

misp attribute.search

Account family: mispApi.

misp attribute.update

Account family: mispApi.

misp event.create

Account family: mispApi.

misp event.delete

Account family: mispApi.

misp event.get

Account family: mispApi.

misp event.getAll

Account family: mispApi.

misp event.publish

Account family: mispApi.

misp event.search

Account family: mispApi.

misp event.unpublish

Account family: mispApi.

misp event.update

Account family: mispApi.

misp eventTag.add

Account family: mispApi.

misp eventTag.remove

Account family: mispApi.

misp feed.create

Account family: mispApi.

misp feed.disable

Account family: mispApi.

misp feed.enable

Account family: mispApi.

misp feed.get

Account family: mispApi.

misp feed.getAll

Account family: mispApi.

misp feed.update

Account family: mispApi.

misp galaxy.delete

Account family: mispApi.

misp galaxy.get

Account family: mispApi.

misp galaxy.getAll

Account family: mispApi.

misp noticelist.get

Account family: mispApi.

misp noticelist.getAll

Account family: mispApi.

misp object.search

Account family: mispApi.

misp organisation.create

Account family: mispApi.

misp organisation.delete

Account family: mispApi.

misp organisation.get

Account family: mispApi.

misp organisation.getAll

Account family: mispApi.

misp organisation.update

Account family: mispApi.

misp tag.create

Account family: mispApi.

misp tag.delete

Account family: mispApi.

misp tag.getAll

Account family: mispApi.

misp tag.update

Account family: mispApi.

misp user.create

Account family: mispApi.

misp user.delete

Account family: mispApi.

misp user.get

Account family: mispApi.

misp user.getAll

Account family: mispApi.

misp user.update

Account family: mispApi.

misp warninglist.get

Account family: mispApi.

misp warninglist.getAll

Account family: mispApi.

SeaTable accounts and staged requests

Save an existing account as seaTableApi. For example:

{
  "profile": "seatable-staged-v2",
  "kind": "api-token",
  "environment": "selfHosted",
  "domain": "https://tables.example.org/team",
  "baseUuid": "11111111-1111-4111-8111-111111111111",
  "providerWorkspaceId": 42,
  "permission": "write",
  "token": "YOUR_EXISTING_TOKEN"
}

Choose cloudHosted with exactly https://cloud.seatable.io, or selfHosted with the exact public HTTPS installation and optional subpath. Enter the selected base UUID, SeaTable workspace ID and read or write permission explicitly. Save/reopen validate locally and leave authentication and permissions unverified. Test, OAuth and automatic refresh are unavailable. A base-token additionally needs an explicit future expiresAt in ISO UTC, within three days. API tokens must omit that field. Never use decoded JWT claims as proof of a grant.

The sixteen actions cover row creation, deletion, retrieval, listing, locking, search, unlocking and updates; base snapshots, metadata and collaborators; adding, listing and removing links; temporary download capabilities and asset upload. Supply existing table/column names and exact 22-character row IDs manually. Columns Ui accepts {"columnValues":[{"columnName":"Enabled","columnValue":false},{"columnName":"Count","columnValue":0},{"columnName":"Title","columnValue":""}]}. These values retain their types when saved. Options accepts only each action's declared keys. Listing/search use an explicit start and a limit of 1–1,000; link lists allow 1–100. Each returns one bounded response, without automatic pages or readbacks.

Every physical stage pauses for a separate signed Human Task assigned to the original run actor, expiring after 15 minutes. Stages can include token issuance, metadata/collaborator reads, the selected operation, file capability issuance, upload and attachment. Review the account, base, method, destination, body and request digest. Current workspace and organization authority, selected account, workflow, input and run must still match. Another Member or Owner cannot approve for the actor. Durable workflow execution, Human Tasks and the SeaTable execution switch must be enabled. API-token exchange explicitly requests a three-day grant; only the selected server's matching response can establish the local grant.

Uploads require an actual owned workspace file with matching content, revision and completed clean scan. Review replacement/append options. An upload can finish before its row attachment is approved; canceling later does not remove the uploaded asset. Temporary download capabilities can disclose file access and are not automatically fetched. Each confirmed stage makes one attempt, including when workflow retries are enabled. An uncertain attempt is not replayed. Inspect the provider before intentionally repeating it; outcomeVerified:false does not verify persistence or delivery. Inputs and selected results can remain in workflow history.

Only ordinary n8n-nodes-base.seaTable imports with explicit typeVersion:2 use this mapping. V1, versionless imports, Tool aliases and triggers need separate migration. Real provider authentication, binary/scan behavior and matching-revision Review walkthroughs remain unverified.

SeaTable field reference

These are the supported native fields. Nested JSON accepts only the declared keys and preserves boolean, number and text values.

seatable row.create

Account family: seaTableApi.

seatable row.remove

Account family: seaTableApi.

seatable row.get

Account family: seaTableApi.

seatable row.list

Account family: seaTableApi.

seatable row.lock

Account family: seaTableApi.

seatable row.search

Account family: seaTableApi.

seatable row.unlock

Account family: seaTableApi.

seatable row.update

Account family: seaTableApi.

seatable base.snapshot

Account family: seaTableApi.

This operation has no input fields.

seatable base.metadata

Account family: seaTableApi.

This operation has no input fields.

seatable base.collaborator

Account family: seaTableApi.

seatable link.add

Account family: seaTableApi.

seatable link.list

Account family: seaTableApi.

seatable link.remove

Account family: seaTableApi.

seatable asset.getPublicURL

Account family: seaTableApi.

seatable asset.upload

Account family: seaTableApi.

Nextcloud accounts, files and exact requests

Select Basic (nextCloudApi) or an existing OAuth token (nextCloudOAuth2Api) explicitly. Basic uses exactly:

{
  "authentication": "accessToken",
  "accountBaseUrl": "https://cloud.example.org/nextcloud",
  "user": "alice",
  "revision": "1",
  "password": "EXISTING_APP_PASSWORD"
}

OAuth uses the same exact HTTPS installation, username and revision with authentication: "oAuth2", token and a future expiresAt in Unix milliseconds; omit password. Saving and reopening remain local and unverified. Test, login, discovery and refresh are unavailable. Native requests currently require a publicly reachable IPv4 HTTPS installation; localhost and private networks are refused. An optional raw webDavUrl must match this account's supported DAV route exactly.

The 17 ordinary choices cover file copy/delete/download/move/share/upload, folder copy/create/delete/list/move/share, and user create/delete/get/list/update. Supply paths and user IDs manually. Paths begin with /; an empty folder-list path selects the root. Copy and move can overwrite; folder deletion is recursive. Shares can expose data or send mail, and user creation can send invitation or reset mail. Keep only fields for the selected operation: share types use the corresponding recipient; public-link passwords are unavailable on other share types. User lists return one bounded response. Return All does not traverse pages or prove completeness; leave Limit empty when it is enabled.

Each physical request requires a signed Human Task for the original run actor, expiring in 15 minutes, with the exact selected account, destination, method, body and workflow inputs. Current workspace and organization permissions must still pass. Another Member or Owner cannot decide for the actor. Durable execution, Human Tasks and the Nextcloud execution switch must be enabled. One claimed attempt is allowed even when node retries are enabled; inspect uncertain provider effects before deliberately repeating them.

Text uploads preserve an explicit empty body. Binary uploads require an owned Workspace File from a supported upstream binding; downloads are persisted through Workspace Files and its configured scanner. File ownership, readiness, bytes, MIME, hash and full revision must remain current through confirmation and OUTPUT. A direct Nextcloud download can feed an upload; arbitrary other-provider binary outputs are unsupported. A scan-state result does not establish the malware protection of a deployment that uses a test scanner. Large text previews show UTF-8 size and hash; the complete value remains bound to the signed request.

Ordinary n8n-nodes-base.nextCloud mappings cover these operations; this does not establish Tool compatibility or full n8n item, error and output parity. Results remain outcomeVerified: false, with bounded response completeness unknown. Actual provider authentication, installed-version behavior, production file scanning and matching-revision Review remain unverified.

Rundeck API 59 accounts and jobs

Use an existing token with the explicit rundeckApi59 family:

{
  "profile": "rundeck-api59-token",
  "apiBaseUrl": "https://rundeck.example.org/automation",
  "token": "EXISTING_API59_TOKEN"
}

Save the exact public HTTPS installation, including an optional subpath, on port 443. Save and reopen remain local and unverified; Test, login, discovery, version detection and refresh are unavailable. Legacy rundeckApi, API 14, insecure TLS and shell argString settings require explicit migration. The installed server's API 59 support and token rights remain unverified.

Choose job.execute or job.getMetadata and supply the exact Job ID manually. Supported IDs use ASCII letters, digits, underscores and hyphens. Execute Arguments retain the original arguments.arguments collection of unique text name/value rows, including empty, Unicode and "0" values. They become JSON options rather than a shell string. An empty or omitted Filter uses the job/server defaults and can select many remote nodes. A job can run commands, affect infrastructure, notify other systems or incur costs.

Both metadata reads and execution need a separate signed 15-minute Human Task decided by the original run actor. The selected account, exact request, workflow, input, run and current workspace and organization authority must still match. Another Member or Owner cannot approve for that actor. Durable execution, Human Tasks and the Rundeck execution switch must be enabled. Each task permits one physical attempt; retries, polling, output fetch, abort requests and returned-link following are unavailable. Before publishing OUTPUT, the workflow checks the current permissions, approved request and recorded result again. A changed approval or an uncertain result stops OUTPUT. An execution receipt is not job completion. Inspect an uncertain remote effect before deliberately repeating it.

Ordinary version-1 n8n-nodes-base.rundeck imports identify these two operations with an explicit API 59 migration notice. Unknown versions and Tool aliases remain unsupported. Results retain outcomeVerified: false; actual provider acceptance, native sockets and matching-revision Review remain unverified. Request and response bounds are local application policy, not vendor limits.

Nextcloud field reference

Only fields for the selected operation are accepted. Nested values retain their original types and supported empty values.

nextcloud file.copy

nextcloud file.delete

nextcloud file.download

nextcloud file.move

nextcloud file.share

nextcloud file.upload

nextcloud folder.copy

nextcloud folder.create

nextcloud folder.delete

nextcloud folder.list

nextcloud folder.move

nextcloud folder.share

nextcloud user.create

nextcloud user.delete

nextcloud user.get

nextcloud user.getAll

nextcloud user.update

Rundeck API 59 field reference

Only fields for the selected operation are accepted. Nested values retain their original types and supported empty values.

rundeck job.execute

rundeck job.getMetadata

Rocket.Chat messages and existing accounts

Save an existing account with the rocketchatRestV1Api family:

{
  "profile": "rocketchat-rest-v1-token",
  "domain": "https://chat.example.org/team",
  "userId": "existing-user-id",
  "authKey": "existing-auth-key"
}

Use the exact public HTTPS installation and optional subpath on port 443. Save and reopen are local and unverified. Test, login, discovery, OAuth and refresh are unavailable. Legacy rocketchatApi accounts require an explicit resave with this profile. Saving does not establish the installed server version or message permissions.

The ordinary version-1 n8n-nodes-base.rocketchat node supports chat.postMessage: one channel or room ID, text, Options and attachments. Channel arrays, Tool and Trigger nodes, missing or unknown versions, threads, uploads and arbitrary custom fields are unsupported. Options keep alias, avatar and emoji. Alias and avatar require the server's message impersonation permission. parseUrls defaults visibly to false; attachment and avatar links may still cause Rocket.Chat servers or clients to fetch remote content.

Choose named attachment fields or JSON Parameters. Named fields preserve supported empty strings, false and 0, including nested fields.fieldsValues. JSON Parameters accepts the supported snake_case attachment array. Switching modes in the editor clears the inactive draft; raw or imported inactive fields are refused. At least nonempty text or a supported nonempty attachment is required. The original node declared messageLink but checked messageLinks during execution. Using named messageLink therefore requires explicit messageLinkMigrationAcknowledged: true to map it to message_link; this correction changes the original execution behavior.

Each message needs a signed 15-minute Human Task decided by the original run actor. Review the selected account, resolved fields, destination and exact request before approving. Current workspace and organization permissions, account, workflow, input, run and node must still match. Another Member or Owner cannot approve for that actor. Durable execution, Human Tasks and the Rocket.Chat execution switch must be enabled. Posting may notify users and disclose message data. Each approval allows one physical attempt; automatic retries, redirects, follow-up reads and returned-link navigation are unavailable. Inspect an uncertain remote effect before deliberately repeating it.

The result keeps the bounded complete success: true response, including the message ID and room ID, with outcomeVerified: false. A response is not proof of message delivery. The validated response is saved as a completed physical attempt before the final OUTPUT authority check. Inputs and responses may persist in history. Actual provider acceptance, native sockets, installed-version compatibility and matching-revision Review remain unverified. Request, response and approval bounds are local application policy, not vendor limits.

Rocket.Chat field reference

rocketchat chat.postMessage

Elasticsearch REST v8 documents and indices

Select the explicit elasticsearchRestV8Basic account family and enter an existing Basic account:

{
  "profile": "elasticsearch-rest-v8-basic",
  "baseUrl": "https://elastic.example.org/search",
  "username": "EXISTING_USERNAME",
  "password": "EXISTING_PASSWORD"
}

Use the exact public HTTPS installation on port 443, including an optional subpath. This profile currently requires printable ASCII Basic account fields and conservative installation paths; these are application restrictions. Saving and reopening remain local and unverified. Test, discovery, version detection, login, refresh, OAuth, API keys and TLS bypass are unavailable. Legacy elasticsearchApi accounts require explicit migration; saving does not establish the installed version or its read, write or delete permissions.

Five document operations are available: document.get, document.index, document.update, document.delete and document.searchPage. Supply one literal Index ID and, where required, Document ID. Index with a Document ID creates or replaces that document; without one, the server assigns an ID. Server defaults may create an index or run ingest processing. Update merges a partial document and can return noop; Delete removes one document. These operations can disclose data, consume resources and incur costs.

Select parameter schema 2, acknowledge the migration, and keep Simple disabled. Ordinary version-1 n8n imports require this explicit migration to raw response namespaces; source fields cannot overwrite response identity. Missing or unknown versions and Tool aliases remain unsupported. Fields to Send retains imported text values; named controls also accept explicit JSON values such as false and 0. Quote those values when text is intended. Auto Map requires an explicit input object and supports Inputs to Ignore. Inactive fields and unsupported nondefault options are refused by name; remove or repair them before running.

Search Page returns one bounded page with a default limit of 50 and a maximum of 100. Its query accepts only match_all, term or match; source include/exclude filters are supported. Partial search results and timed-out or failed-shard responses are refused. Raw hit metadata and total relation remain available; this is not an all-results export. Bulk, arbitrary query DSL, scripts, upsert, aliases or multiple target syntax, cross-cluster targets, PIT, pagination, refresh and pipelines are outside this implemented subset. A literal name can resolve to an alias on the server; Save does not discover its target, and a returned index identity mismatch is refused.

Each physical read or write needs its own signed 15-minute Human Task approved by the original run actor. The selected account, exact request, workflow, run, input, version, live switch and current workspace and organization authority must still match. Another Member or Owner cannot approve for that actor. One physical attempt is allowed, including when node retries are configured. A complete response is recorded before OUTPUT, which checks current authority and the exact stored result again. An uncertain result is not retried or checked with a hidden follow-up request; inspect remote effects before deliberately repeating it.

All results retain outcomeVerified: false. Successful local simulation does not establish provider acceptance, native DNS/TLS sockets, installed server behavior, real account permissions or matching-revision Review behavior. Request, response, time and structural bounds are application policy, not Elasticsearch vendor limits.

Index operations

Four index operations use the same account and explicit schema-2 migration:

Create, Delete and Get require one literal Index ID. Wildcards, _all, multiple targets, cross-cluster syntax and path expressions are refused. A literal name can still resolve to an alias on the server; saving does not discover its target.

Index Get All defaults to 50 returned namespaces and allows at most 100. The complete response must fit the local 2 MiB budget and pass validation before slicing. The result keeps the selected raw namespaces and coverage.examined, returned, truncated and limit; allResultsComplete and remoteLimitApplied remain false. A short or empty result does not establish a complete inventory. Returned names may include data streams.

Index Create accepts the original empty aliases, mappings and settings defaults and the original default timeout, shard-wait and include_type_name values only as omitted controls. Index Get likewise accepts its original default additional fields only as omitted controls. Server defaults apply and may differ from those original values. Nondefault or unknown fields are refused by name; aliases, mappings, settings and query options are not transmitted. Delete and Get All accept no additional-field or option entries.

False, missing or malformed acknowledgements leave the remote outcome uncertain. Every index read or write still requires the original actor's separately signed request, current workspace and organization authority, one physical attempt and a checked stored result before OUTPUT. An uncertain outcome never triggers an automatic retry, readback or compensating request. Results retain outcomeVerified: false; local simulation does not verify installed-server permissions or remote effects.

Elasticsearch document field reference

elasticsearch document.get

elasticsearch document.index

elasticsearch document.update

elasticsearch document.delete

elasticsearch document.searchPage

Elasticsearch index field reference

elasticsearch index.create

elasticsearch index.delete

elasticsearch index.get

elasticsearch index.getAll

ServiceNow incident requests

The ordinary version-1 n8n-nodes-base.serviceNow integration supports five incident operations with explicit Basic authentication and parameter schema 2. Acknowledge migration and keep Simple disabled. Other resources, attachments, OAuth, Tool/Trigger aliases and missing or unknown versions remain unsupported.

Save an existing serviceNowTableV2Basic account using named fields or Advanced JSON:

{
  "profile": "servicenow-table-v2-basic",
  "baseUrl": "https://example.service-now.com",
  "username": "EXISTING_USERNAME",
  "password": "EXISTING_PASSWORD"
}

Use an exact public HTTPS service-now.com tenant origin on port 443, with one subdomain label and no path. Save and reopen are local, make zero provider requests and leave authentication unverified. Test, login, discovery and OAuth are unavailable. Legacy Basic accounts require explicit migration and reselection; saving does not establish tenant version, roles or ACL permissions.

IDs require exactly 32 lowercase hexadecimal characters. Short description is nonempty and limited locally to 160 UTF-8 bytes; Description allows an explicit empty string and at most 4096 characters. Impact and Urgency accept integers 1–3. Unknown and inactive fields are refused by name. Read options allow distinct sys_id, number, short_description, description, impact, urgency and active projections and must include sys_id; display values false, true or all and Exclude reference link are optional. Keep boolean false distinct from an omitted field.

Writes may trigger business rules, notifications and downstream workflows. Lists retain unknown completeness even when short or empty: server limits and ACL filtering do not establish a complete inventory. Raw incident text and responses may persist in run history or reach downstream AI.

Each physical read or write requires its own signed 15-minute Human Task decided by the original run actor. The exact saved account, request, workflow, run, input, version, live switch and current workspace and organization permissions must still match. Another Member or Owner cannot approve for that actor. Durable Human Tasks and the ServiceNow execution switch must be enabled. One physical attempt is allowed, even with node retries; an uncertain outcome is never retried or checked by a hidden readback. Final OUTPUT rechecks authority and the stored result. Results keep outcomeVerified: false. Local simulated transport tests do not verify a real tenant, native sockets or matching-revision Review behavior. Bounds are application policy, not vendor limits.

Gong call reads

The ordinary version-1 n8n-nodes-base.gong integration supports call Get and one-page List through explicit schema-2 migration. Transcript is a separate native read, with its own approval; selecting transcript inside an imported Get does not cause a hidden request. Acknowledge the raw-response migration and select the matching saved account. Tool aliases, unknown versions, automatic cursors, media retrieval, users and OAuth are outside this implemented subset.

Save an existing gongRestV2Basic account with Access Key and Access Key Secret:

{
  "profile": "gong-rest-v2-basic",
  "baseUrl": "https://api.gong.io",
  "accessKey": "EXISTING_ACCESS_KEY",
  "accessKeySecret": "EXISTING_ACCESS_KEY_SECRET"
}

This profile supports only the explicitly selected https://api.gong.io origin on port 443. It does not discover the customer's region, tenant, plan, scopes, key expiry or trusted-IP policy. Legacy gongApi and OAuth accounts need explicit migration and reselection. Save and reopen are local and unverified, with zero provider requests; Test, discovery, login and refresh are unavailable.

Use exact decimal ID strings of 1–20 digits, preserving leading zeros; numeric IDs and automatic locator searches are refused. List filters allow at most five distinct Call IDs or Primary user IDs. Times use UTC seconds, for example 2026-10-21T10:00:00Z, with From before To. Get supports Include fields for parties, topics, brief, public comments, highlights, key points, call outcome, outline, structure and trackers; List supports parties and topics. Named and JSON controls preserve explicit false. Unsupported fields are refused rather than silently dropped.

All three POST requests are reads. The complete bounded raw page can contain participants, public comments, analysis or transcripts, including records beyond the local selected-item limit. Those raw namespaces may persist or reach downstream AI; a local limit is not a server-side privacy limit or a complete export. The local response budget is 1 MiB, with a 30-second transport bound; these are application restrictions.

Each read requires the original actor's separately signed 15-minute Human Task, the exact selected account and request, and current workspace and organization authority. Durable Human Tasks and the Gong execution switch must be enabled. One physical attempt is allowed, including with node retries. Settlement and final OUTPUT authenticate the stored response and receipt; an uncertain read is not retried, refreshed or followed by a hidden request. Results keep outcomeVerified: false. Provider acceptance, installed permissions, native sockets and matching-revision Review remain unverified.

Okta user requests

The ordinary version-1 n8n-nodes-base.okta integration supports five user operations through explicit parameter schema 2 and manual SSWS authentication. Acknowledge migration and select the matching saved account. OAuth, unknown versions, Tool/Trigger aliases, password changes, lifecycle operations and custom profile attributes are outside this implemented subset.

Save an existing oktaSswsUsersV1 account using named fields or Advanced JSON:

{ "profile": "okta-ssws-manual-v1", "origin": "https://example.okta.com", "accessToken": "EXISTING_SSWS_TOKEN" }

Use the exact public HTTPS tenant origin on port 443, without a trailing slash, path, query or fragment. Save and reopen are local and leave authentication unverified, with zero provider requests. Test, login, OAuth, discovery and refresh are unavailable. Legacy accounts require explicit migration and reselection; saving does not establish token permissions or tenant policies.

Manual user IDs start with 00u and use the supported ASCII form; me, login lookup and locator searches are refused. Profile controls support the 31 named standard fields, including Postal address and Zip code. Required First name and Last name retain the API's explicit null option. Null, empty text and omitted fields remain distinct where supported. Duplicate create fields, unknown attributes and expressions are refused. A Link header is untrusted data: the application never follows it automatically. A short page is not a complete inventory.

Raw user profiles may persist in run history or reach downstream AI. Each physical request requires the original actor's separately signed 15-minute Human Task, the exact selected account and request, and current workspace and organization permissions. Another Member or Owner cannot approve for that actor. Durable Human Tasks and the Okta execution switch must be enabled. There is one physical attempt, including with node retries; uncertain outcomes do not cause a hidden readback or a second delete. Final OUTPUT authenticates the stored response and receipt and rechecks authority. Results keep outcomeVerified: false. The local response budget is 1 MiB with a 30-second transport bound. Simulated tests do not verify a real tenant, provider permissions, native sockets or matching-revision Review behavior.

Unleashed sales orders and stock

The ordinary version-1 n8n-nodes-base.unleashedSoftware integration supports one sales-order page, one product's stock and one stock page through explicit parameter schema 2. Acknowledge migration and select the matching company account. Automatic pagination, writes, unknown versions, Tool/Trigger aliases and unsupported filter values are outside this implemented subset.

Save an existing unleashedHmacCompanyV1 account using named fields or Advanced JSON:

{
  "profile": "unleashed-hmac-sha256-query-manual-v1",
  "apiId": "EXISTING_API_ID",
  "apiKey": "EXISTING_API_KEY",
  "clientType": "confluye/local"
}

The origin is fixed to https://api.unleashedsoftware.com on port 443. Client type uses two lowercase letter, digit or hyphen segments separated by /, with at most 64 characters. The server signs the exact outgoing query with the company's API key. Save and reopen are local and unverified, with zero provider requests; Test, discovery, login, OAuth and refresh are unavailable. Saving does not verify company access, subscription or permissions.

Pages default to 1 and allow 1–1,000,000; limits default to 100 and allow 1–100. These are local restrictions. Keep Return all disabled. Sales filters include Customer ID GUID lists, Customer code, End date, Modified since, Order number, Order status and Start date. An Order number cannot be combined with other filters because the provider overrides them. Status supports Backordered, Completed, Deleted, Parked and Placed; omitting it does not include deleted orders automatically.

Stock filters include As-at date, Is assembled, Modified since, Product ID GUID lists, Warehouse code and Warehouse name. Explicit false is preserved. Order by supports only the empty default until a closed column contract is available. GUID lists preserve order and duplicates and allow up to 50 entries. Dates require real YYYY-MM-DD values; imported date-time truncation is refused. Raw .NET date representations remain raw, and a missing stock record does not establish zero stock.

Customer, order, stock and cost data may persist or reach downstream AI. Each read requires the original actor's separately signed 15-minute Human Task, the exact company account and request, and current workspace and organization authority. Durable Human Tasks and the Unleashed execution switch must be enabled. One physical attempt is allowed, including with node retries; there is no automatic next-page, count, retry or readback request. Settlement and final OUTPUT authenticate the stored response and receipt. Results keep outcomeVerified: false. The local response budget is 1 MiB with a 30-second transport bound. Provider acceptance, native sockets, company permissions and matching-revision Review remain unverified.

AWS SQS and Lambda requests

AWS SQS supports one sqs.sendMessage request; AWS Lambda supports one lambda.invoke request. Ordinary version-1 imports require explicit parameter schema 2 and migration acknowledgement. These selected actions do not provide complete AWS credential, region, binary or node parity.

Save an existing awsSqsManualSigV4V1 or awsLambdaManualSigV4V1 account using named fields or Advanced JSON. Use the matching profile, aws-sqs-manual-sigv4-v1 or aws-lambda-manual-sigv4-v1, partition aws, a supported region, the declared 12-digit AWS account ID, Access key ID and Secret access key. Choose manual-static, or manual-temporary with an existing Session token and explicit UTC expiresAt. Temporary credentials must remain valid beyond the 30-second request budget plus a 60-second safety margin; replace and approve them again when necessary. The declared account is an input restriction, not a verified AWS identity.

Save and reopen send zero provider requests and leave authentication unverified. There is no connection Test, sign-in, STS, role assumption, system credential chain or refresh. The fixed endpoint map includes 34 commercial regions per service. GovCloud, China, isolated partitions, custom, FIPS and dual-stack endpoints are unsupported; the map does not promise every current AWS region.

SQS supports the selected string and canonical decimal number attributes, preserves omitted delay versus zero and validates returned body/attribute MD5 values. FIFO requires explicit group and deduplication IDs and refuses per-message delay, including zero. Binary attributes are refused. Lambda does not accept an empty payload, binary results, logs or client context. Function versions, aliases and $LATEST can change without a configuration read. Messages and invocations may incur charges.

Every request needs the original actor's separately signed 15-minute approval, exact saved account and request, and current workspace and organization authority. Durable Human Tasks and the service execution switch must be enabled. SigV4 date and signature are generated at physical dispatch; approval binds the stable request and saved account revision. There is one physical attempt, with no automatic resend or readback after an uncertain result. Settlement and final OUTPUT check the authenticated stored response receipt. outcomeVerified: false does not establish downstream completion.

Business inputs and bounded JSON responses can persist in workflow history and reach downstream AI. Local limits are 32 KiB per request, 1 MiB per response and 30 seconds; these are application restrictions, not AWS limits. Local simulated tests do not establish actual AWS permissions, endpoint acceptance, native sockets or matching-revision Review behavior.

ServiceNow fixed-table ID lookups

Alongside the five incident actions, ServiceNow supports six first-page ID lookups using the same serviceNowTableV2Basic account and explicit schema-2 migration:

Explicitly enable lookupIdProjectionAcknowledged. This migrates the original Get all action to a restricted first page of sys_id values only. The limit is 1–50, default 50, with offset zero, display values disabled and reference links excluded. There is no custom query, arbitrary table, count request or automatic pagination. Empty and short pages do not establish a complete inventory. IDs can still disclose organizational structure and are retained under ordinary history and downstream-data rules.

Only HTTP 200 JSON with the exact result envelope and rows containing one lowercase 32-hex sys_id is accepted. Extra fields and display data are refused before public output. Each lookup requires the original actor's signed approval, current workspace and organization authority and one physical claim. Aborted or uncertain lookup requests settle without automatic replay; inspect the provider separately before deciding to repeat. Attachments, arbitrary table-record actions, user actions and OAuth are outside this subset. These six lookups have local simulated coverage; real provider permissions, native sockets and matching-revision Review remain unverified.

Twist text and JSON actions

The ordinary version-1 Twist integration supports 22 selected text/JSON actions through explicit parameter schema 2 and migration acknowledgement:

Save a twistManualBearerV1 account using named fields or Advanced JSON:

{ "profile": "twist-manual-bearer-v1", "accessToken": "EXISTING_MANUAL_TOKEN" }

This is a partial manual-token authentication profile. The captured official documentation provides testing tokens and requires OAuth for production public integrations. Original OAuth credentials do not migrate automatically. Saving and reopening are local and unverified, with zero provider requests; Test, sign-in, discovery, OAuth acquisition and refresh are unavailable. The endpoint is fixed to https://api.twist.com/api/v3 on port 443. Saving does not establish token identity, workspace membership or permissions.

Use exact parent/workspace/recipient IDs and literal text or structured values. Expressions are unsupported. Supported controls preserve omitted fields, empty text, false and zero where permitted. Timestamps require explicit Unix-second integers; timezone-dependent date conversion is not reproduced. Channel Update requires a nonempty Name. Temporary IDs must be negative; an imported default of zero requires repair. Original uppercase ordering requires lowercase asc or desc, and the deprecated is_starred filter requires an explicit current boolean. Conversation-list additional fields must be moved to native Filters. Conversation Create retains the original workspace ID although the current endpoint documentation does not list it; provider acceptance remains unverified.

Lists require singlePageRawAcknowledged: true; keep Return all disabled. Limits default to 50 and allow 1–500. Channel List receives and examines the complete bounded server array before local slicing; the local limit does not prevent other received records from entering raw history. Other lists make one bounded request. No list proves complete history or follows another page. Comment Get requires explicitly choosing the documented object or original comment-wrapper response shape. Raw HTTP bytes and the native data projection remain separately available; bodyless 200/204 responses and outcomeVerified: false do not prove an external effect.

Creating, editing, deleting, archiving or unarchiving content can change access or notify people. Exact recipients and parent IDs matter; raw user/message data can persist in history or reach downstream AI. Every read and write requires the original actor's separately signed 15-minute approval, the exact saved account and request, and current workspace and organization authority. Durable Human Tasks and the Twist execution switch must be enabled. One physical attempt is allowed, with no automatic retry, reconnect, cursor following or hidden readback after uncertainty. Final OUTPUT validates the signed stored response receipt and current authority.

Attachments and binary input for comment, conversation-message and thread Create/Update are unavailable. Nonempty binary references and attachment arrays are refused; no URL is fetched as a substitute. Action-button links remain provider payloads and are not fetched by this integration. Local limits are 64 KiB request, 1 MiB response and 30 seconds, not provider limits. Local simulated checks do not verify production OAuth, provider DNS/TLS/HTTP, actual encrypted database Save/reopen, the full persisted original-actor approval flow, human MFA or matching-revision Review.

SyncroMSP REST: explicit migration and one approved request

SyncroMSP supports 20 ordinary actions: create, delete, get, update and one-page list for contacts, customers and tickets, plus create, delete, get, mute and one-page list for RMM alerts. Select an existing syncromspBearerRestV1 account with profile syncromsp-bearer-manual-v1, a lower-case ASCII tenant label and an existing bearer token. The destination is fixed to https://tenant.syncromsp.com/api/v1 for that label. Saving and reopening the account is local and unverified; it sends no provider request. Test, sign-in, OAuth, token refresh and automatic discovery are unavailable. Authentication, tenant identity, permissions and provider acceptance still need verification.

Original numeric n8n node version 1 imports preserve the original parameters for repair. They require manual schema-2 conversion, a separately selected supported account, wireMigrationAcknowledged:true and the account's queryAuthMigrationAcknowledged:true: the original query API key is not silently reused as a bearer token. Tool and Trigger identities and other node versions are unavailable. Expressions and unsupported address-array writes remain visible for explicit repair and cannot execute as supported fields.

Review the exact native field names. Customer name and address aliases use their explicit wire names; ticket fields retain explicit empty, null, false and zero values where the schema permits them. An omitted invoice-CC field stays omitted, while an explicit empty list clears it. Updates require at least one selected field. For example, an explicitly migrated customer update can preserve these distinct values:

{
  "parameterSchemaVersion": 2,
  "wireMigrationAcknowledged": true,
  "id": 17,
  "fields": {
    "notes": "",
    "notification_email": null,
    "get_sms": false,
    "invoice_cc_emails": []
  }
}

RMM mute additionally requires muteQueryMigrationAcknowledged:true; it sends the duration in the documented query location rather than the original request body. The retained flat RMM-create body and ticket asset_id/ticket_type fields differ from current documentation; their live compatibility remains unverified. No alternate request is sent automatically to resolve those differences.

Lists require singlePageAcknowledged:true, default to a local limit of 25 and allow at most 100 returned items. The bounded raw page is retained before local slicing; examined and returned counts do not establish completeness. returnAll:true, cursor/Link following and hidden readback are unavailable. Local limits are 32 KiB per request, 1 MiB per response and 30 seconds per invocation. These are application limits, not provider guarantees; IPv6-only destinations are unsupported.

Every read and write needs the original run actor's separate exact-request confirmation and current workspace and organization authority. Approval expires after 15 minutes and is tied to the selected account, workflow, inputs and request. Changing the saved account requires new consent. Contact, customer, ticket and alert data can remain in execution history or reach downstream AI. Review the destination, record IDs, literal body and applicable changes before confirming. One physical attempt is allowed; ambiguous results are not retried automatically. Bodyless HTTP 200/204 and returned IDs do not verify the remote effect; results retain outcomeVerified:false.

Local controls, encryption, import round trips and the production gateway have been exercised with simulated storage and provider transport. Matching Review walkthroughs, current composed database durability, native provider sockets, real authentication/effects and AI-provider interpretation are not established by those local checks.

Taiga: manual token and explicit field migration

Taiga supports create, delete, get, update and one-page list for epics, issues, tasks and user stories. Select taigaBearerManualV1 with profile taiga-bearer-manual-v1, an exact public HTTPS origin and an existing accessToken. Named fields and Advanced JSON use the same encrypted account format. Saving and reopening are local and unverified. The original username/password login requires explicit migration and account reselection; no login, refresh, credential Test or option discovery occurs.

Original ordinary imports retain their type, version, parameters and credential references for repair. Native requests require parameterSchemaVersion:2 and migrationAcknowledged:true. Tool and Trigger identities are separate. Enter IDs manually. Updates require the current record version and manualVersionAcknowledged:true; version 0 is valid and no version lookup runs. Selected assigned_to or milestone null values clear the field where supported, while omission leaves it unchanged. False, empty descriptions or blocked notes and empty tag lists retain their distinct values. Corrected filter names, comma-separated query tags and hash-prefixed epic colors need explicit migration. Legacy user-story type additionally needs unverifiedLegacyTypeAcknowledged:true; live support remains unverified.

Lists need singlePageRawAcknowledged:true, default to a local limit of 50 and allow at most 100 selected items. The bounded raw page remains available; completeness is unknown. Return all, automatic continuation and following server next URLs are unavailable. Local limits are 64 KiB per request, 1 MiB per response and 30 seconds per invocation; these are application limits rather than provider guarantees.

Twake: historical n8n plugin message

Twake supports the ordinary message.send action through twakePluginWorkspaceKeyV1, using profile twake-plugin-workspace-key-v1 and an existing workspaceKey. The fixed destination is https://plugins.twake.app/plugins/n8n/actions/message/save. Basic, JWT and Twake Server accounts are separate unsupported modes. Saving is local and unverified; the historical plugin's current availability, permissions, response status and wrapper have not been verified with a live account.

Enter the channel ID and formatted content, with explicit schema-2 migration. The request sets object.channel_id, object.content.formatted and object.hidden_data.allow_delete:"everyone". Nonempty sender name or icon overrides are forwarded; empty overrides are omitted. An icon URL is forwarded as data and never fetched locally. One native invocation produces one request and one result, explicitly replacing the original multiple-input last-result behavior.

Both providers require the original run actor's separate exact-request confirmation, valid for 15 minutes, plus current workspace and organization authority. Changing the saved account, ciphertext, workflow, inputs or request requires fresh consent. Raw descriptions, messages and personal data can remain in history or reach downstream AI; remote changes can notify recipients or incur costs. One physical attempt is allowed. Ambiguous results are not automatically resent or read back. A response, returned ID or bodyless deletion does not verify the remote effect; outcomeVerified remains false.

The result envelope retains the bounded raw UTF-8 body, digest and item counts. Contract checks require Taiga create 201, delete 204 with no body, other Taiga responses 200 and Twake send 200; these expectations are not live-provider acceptance evidence. Invalid encoding, leading BOM, duplicate JSON keys, oversized bodies and reflected account secrets are refused. Local storage, request, approval and runtime tests use simulated database facts and provider transport. Current composed PostgreSQL durability for these providers, native sockets, real authentication/effects, AI interpretation and matching Review walkthroughs remain untested.

Telegram and KoBo ordinary actions

Telegram exposes 23 synchronous Bot API actions; KoBo exposes 17 JSON/metadata actions. Select the exact supported saved account and repair imported schema 2 controls explicitly. Original versions, parameters and credential references remain available for repair. These action counts exclude Tool/HitlTool/Trigger identities, binary upload/download and Telegram sendAndWait.

Telegram file metadata requires download:false; single-media actions require binaryData:false and preserve bot file IDs or provider-fetched URLs. KoBo file metadata also requires download:false; URL-reference creation sends metadata without downloading it in the application. KoBo lists require a single-page acknowledgment and explicit limit; next/count never establish completeness. Typed false, zero, null and empty values follow each action's declared controls. See Telegram fields and limits and KoBo fields and limits.

Review the exact request before the original actor approves it. Every invocation makes one physical attempt with current workspace and organization authority; an uncertain result never resends automatically. Raw retained responses can enter execution history and downstream AI. HTTP acceptance is not delivery, deletion or remote-effect verification.

AWS Comprehend, SNS and SES requests

AWS Comprehend supports three text analyses; SNS supports topic create, delete and publish; SES supports six custom verification email actions, two email sends and five template actions. These nineteen ordinary version-1 imports require parameter schema 2, explicit migration acknowledgement and an acknowledgement that inputs and operation data persist in workflow history and may reach downstream AI or remote actions. The SNS Trigger lifecycle, Tool variants, extra AWS APIs, attachments and raw MIME are unsupported.

Select the exact manual credential family: awsComprehendManualSigV4V1, awsSnsManualSigV4V1 or awsSesManualSigV4V1. Their profiles are respectively aws-comprehend-manual-sigv4-v1, aws-sns-manual-sigv4-v1 and aws-ses-manual-sigv4-v1. Named fields and Advanced JSON save the same strict account. Supply partition aws, an explicitly supported commercial region, a declared 12-digit account ID, mode, access key ID and secret access key. manual-temporary additionally requires a supplied session token and UTC expiry beyond the 30-second invocation budget plus a 60-second margin. manual-static forbids token and expiry. Save/reopen is local and unverified, performs zero provider HTTP and does not reveal saved secrets. Test is unsupported and bulk Test skips these three families. No STS, role assumption, system provider chain, refresh, account probe, discovery or picker read occurs. The account ID restricts inputs and does not prove AWS identity.

Each run requires a separate exact-request HumanTask approved by the original run actor within fifteen minutes. Another administrator cannot substitute. Current workspace and organization membership, policy, workflow/node/input/account revisions, preference generation, consumed task/wait and lease are checked again at execution and OUTPUT. One checkpoint claim allows at most one physical request. An error, timeout, disconnect, lost settlement or missing receipt leaves an uncertain outcome and never grants automatic resend. HTTP acceptance, MessageId or TopicArn does not prove delivery, processing, verification, topic deletion, remote template state, privileges or identity. Comprehend discloses text; SNS can fan out to existing subscribers; SES may send mail or change remote templates. Charges and downstream effects are possible.

Exact bounded raw response bytes are application-encrypted into WorkspaceFiles and bound to the run, actor, account, request, task/wait and HTTP status. OUTPUT reopens and authenticates that protected artifact under current authority. This does not claim raw database encryption, inspection of plaintext by antivirus, or completed backup/export deletion verification. The configured file scanner sees ciphertext; unavailable storage/scanning prevents a usable result without resending. Operation data, including addresses, templates and other personal information, remains in ordinary workflow history and can reach downstream AI and remote actions. Require rawRetentionAcknowledged:true before execution; do not supply data you cannot retain or disclose through those paths.

These profiles impose local limits below provider limits: 4096 UTF-8 bytes per text/template field, 16 KiB request, 1 MiB response, thirty seconds, sixty-four template-data rows and fifty total destination recipients. Lists are exactly one bounded page, default twenty, at most fifty custom verification templates or one hundred ordinary templates. Require boundedPageAcknowledged:true, returnAll:false and explicit opaque continuationToken for a subsequent separately approved page. No hidden pagination, readback or retry occurs. HTML, redirect and template strings remain inert locally and are never fetched. The supported endpoint maps contain 12 Comprehend, 34 SNS and 27 SES commercial regions. Available regions depend on the selected service.

SNS names and ARNs are bound to the selected commercial region and declared account. FIFO create/delete accepts .fifo; FIFO publish is refused because the ordinary source lacks MessageGroupId. Comprehend retains ordered language/entity arrays, zero scores and numeric spelling in protected raw data; dominant-language simple mode also provides a derived map. SES custom verification update repairs failureRedirectionURL and fromEmailAddress; template HtmlPart is controlled by its own presence, including HTML-only updates. Unknown/hidden original values and original credential references remain in the import source ledger until deliberately repaired. Corrected form encoding replaces the legacy raw query concatenation and JSON-quoted form body. Import, repair, serialize, reopen and reimport do not grant account or provider verification.

The exact ordinary operation IDs are:

• sns.create
• sns.delete
• sns.publish
• comprehend.text.detectDominantLanguage
• comprehend.text.detectSentiment
• comprehend.text.detectEntities
• ses.customVerificationEmail.create
• ses.customVerificationEmail.delete
• ses.customVerificationEmail.get
• ses.customVerificationEmail.getAll
• ses.customVerificationEmail.send
• ses.customVerificationEmail.update
• ses.email.send
• ses.email.sendTemplate
• ses.template.create
• ses.template.delete
• ses.template.get
• ses.template.getAll
• ses.template.update

For example, an HTML-only SES template update uses:

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "templateName": "existing-template",
  "updateFields": { "htmlPart": "<p>Reviewed replacement content</p>" }
}

HaloPSA client, site, ticket and user actions

The halopsa provider offers the following twenty ordinary operation IDs:

Every action requires a selected existing manual account, schema-2 migration and original-actor approval. For example, an explicitly reviewed client create uses:

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "onePhysicalAttempt": true,
  "clientName": "Example customer",
  "additionalFields": { "is_vip": false, "notes": "" }
}

A client list uses one explicit page:

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "onePhysicalAttempt": true,
  "singlePageAcknowledged": true,
  "page_no": 1,
  "page_size": 50,
  "returnAll": false
}

getAll does not mean a complete collection. limit, when supplied, must equal page_size. A different page needs its own approval. Optional selected field controls retain admitted null, false, zero and empty values; exact schemas still reject incompatible types and inactive fields. User-password input accepts only saved-account-password and uses the selected account's encrypted userPassword. Public outputs and exports omit protected request/response content. See setup, retention and effect limits.

GoToWebinar ordinary actions

Choose a saved account and review the unchecked schema-2 acknowledgements. A webinar read uses:

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "webinarKey": "9007199254740993"
}

A registrant list reads exactly one page:

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "singlePageAcknowledged": true,
  "webinarKey": "200",
  "page": 0,
  "limit": 50,
  "returnAll": false
}

Enter resource, question and answer IDs manually. Named nested controls and Advanced JSON preserve admitted false/empty values and exact int64 lexemes. Source aliases include surveyAnswers → surveys, timezone → timeZone, isOnDemand → isOndemand, and AnswerKey → answerKey. Attendee details cover polls/questions/surveys; session details also include performance. Date ranges and session times must be explicit. Coorganizer creation distinguishes an internal organizer key from external given-name/email input. Registrant creation preserves address, optional fields and simple/multiple-choice answer collections. Reinvitations are bodyless POST requests.

webinar.update requires explicit notifyParticipants, complete updateFields.times and completeSessionDefinitionAcknowledged:true. single_session/series work; source sequence is refused. Read response privacy, downstream limits and possible effects before confirmation.

Clearbit and Medium ordinary actions

All five use parameterSchemaVersion:2, migrationAcknowledged:true, onePhysicalAttempt:true and rawResponseRetentionAcknowledged:true; the controls begin unchecked. For autocomplete, select the public policy account and enter:

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "onePhysicalAttempt": true,
  "rawResponseRetentionAcknowledged": true,
  "name": "Example"
}

Medium post creation also requires sourceEnvelopeMigrationAcknowledged:true and tagLimitMigrationAcknowledged:true. Choose html or markdown, and an explicit public, draft or unlisted publish status. notifyFollowers:false stays false on the wire. Tags are trimmed and limited to three nonblank entries, without silently truncating the source's five; each has separate 25-codepoint, 25-UTF-16-unit and 100-byte bounds. Titles have separate 100-codepoint and 100-UTF-16-unit bounds. HTML and canonical URLs are inert submitted content, never application-executed or fetched.

Clearbit hints preserve explicit empty strings; null, false, zero and non-string hints refuse. Company aliases include companyName → company_name; person aliases include givenName, familyName, ipAddress, companyDomain and exact linkedIn → their snake-case/lowercase wire names. Hints are allocated afresh for each request. Medium schema 2 unwraps data for both post destinations while retaining raw response and source-item envelope ledgers. Publication lists retain one actual response, order and duplicates, with no hidden /me, automatic paging or local slicing. See existing-access prerequisites and data retention.

FileMaker records, find and scripts

All actions require parameterSchemaVersion:2 and initially unset migrationAcknowledged, rawRetentionAcknowledged, manualSessionAcknowledged and onePhysicalAttempt, explicitly chosen true. Enter layoutSchema.fields and layoutSchema.portals manually; field types are string/number/boolean with optional nullable declaration. Unknown fields, alternate response layouts, portal writes and automatic schema discovery are unsupported. Create/edit send empty portal data. records and find also require singlePageAcknowledged, explicit offset at least 1 and limit 1–100. There is no automatic pagination.

A completed Get record example, whose acknowledgments are shown explicitly true rather than UI defaults:

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "onePhysicalAttempt": true,
  "manualSessionAcknowledged": true,
  "layout": "Customers",
  "recid": "17",
  "getPortals": false,
  "layoutSchema": {
    "fields": [
      {
        "name": "Name",
        "type": "string"
      }
    ],
    "portals": []
  }
}

getPortals:false sends an explicit empty portal list; it does not omit the API field and retrieve every portal. Source sort rules become the records _sort JSON query; find includes selected portals, after-script uses its actual scriptAfterParam, and duplicate scripts use the JSON body. These are disclosed corrections to source behavior. Null, false, zero and empty strings stay distinct; disabled/unknown controls, unsafe numbers and unresolved expressions require manual repair. Returned field/portal data must match the supplied schema. Useful business data and raw/source envelopes remain available for downstream mapping and can persist in ordinary history or downstream AI. See existing-session setup.

Bitwarden organization management actions

All actions require parameterSchemaVersion:2, migrationAcknowledged, sessionScopeAcknowledged, rawRetentionAcknowledged and oneAttemptAcknowledged, initially unset and deliberately chosen true. Enter exact UUIDs manually; no dynamic discovery runs. Lists additionally require singlePageAcknowledged and pageMaxItems 1–1000; optional returnAll can only be false. The cardinality cap is local refusal, not a provider limit or local slice. Events require explicit start/end ISO timestamps, end after start; optional acting user, original itemID and continuation token are entered manually. The original itemID maps to current itemId. A next event page requires separate confirmation.

Deletes and membership replacement are consequential writes. group.updateMembers replaces the exact unique memberIds array, and member.updateGroups replaces the exact unique groupIds array, at most 1000 UUIDs. Empty arrays clear all associations. Both membership replacement and consequential-effect acknowledgments are required for replacements; deletes require the latter. There is no hidden readback to reconstruct membership.

This completed group-membership clear example shows explicit acknowledgments, never UI defaults. It may remove access and must receive its own approval:

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "sessionScopeAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "oneAttemptAcknowledged": true,
  "groupId": "22222222-2222-4222-8222-222222222222",
  "memberIds": [],
  "membershipReplacementAcknowledged": true,
  "consequentialEffectAcknowledged": true
}

Current collection/group/member models admit documented lowercase type, readOnly, permissions and nullable business fields. Unknown schema/default/nullability refuses. Business data, including PII and raw/source JSON, can persist in ordinary history/checkpoints and reach downstream mapping or AI; there is no new raw-history encryption, TTL, scanner or deletion-completion guarantee. Local bounds include 64KiB input/request, 1MiB response and a 30-second invocation, with stricter signed-approval/output limits where applicable. See session setup.

Cortex jobs and reports

Ordinary n8n-nodes-base.cortex, numeric typeVersion 1, exposes exactly job.get (Get job, cortex_job_get) and job.report (Get job report, cortex_job_report). Choose the saved Cortex key account, manually enter jobId using 1–256 ASCII letters/digits/underscore/hyphen, and explicitly set parameterSchemaVersion:2, migrationAcknowledged, rawRetentionAcknowledged and onePhysicalAttemptAcknowledged. All acknowledgments start unset. Named and JSON controls share the same strict materializer.

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "onePhysicalAttemptAcknowledged": true,
  "jobId": "job_42"
}

This sends exactly one GET to the saved base path plus /api/job/job_42 or /api/job/job_42/report. Job result.data preserves actual id/status/dataType and bounded authorized business fields. Report summary/full/success/artifacts remain useful business JSON, without inventing a universal analyzer schema. A pending report stays a status string; there is no polling or waiting for completion. Map node.<id>.json.data.id or the actual returned report field downstream. Bounded rawBodyUtf8 and sourceItemsJson are ordinary retained data, not protected-file references. No returned URL is fetched.

n8n software Public API actions

Ordinary n8n-nodes-base.n8n, numeric typeVersion 1, exposes these nine operations:

All nine require parameterSchemaVersion:2, initially unset migrationAcknowledged, rawRetentionAcknowledged and oneAttemptAcknowledged, plus a separate exact-request decision. Lists additionally require unchecked singlePageAcknowledged, explicit returnAll:false, integer limit 1–250 and optional literal cursor. Execution metadata reads (execution.get and execution.getAll) require options.activeWorkflows:false, which materializes includeData=false; execution deletion accepts its explicit ID and destructive acknowledgment without read options; runData/workflowData/customData are omitted. Execution IDs are positive decimal strings within the safe-integer bound; workflow IDs and credential type names use the disclosed literal ASCII letters/digits/underscore/hyphen subset. Response IDs remain strings, never coerced numbers.

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "oneAttemptAcknowledged": true,
  "executionId": "42",
  "options": { "activeWorkflows": false }
}

A workflow page can use singlePageAcknowledged:true, returnAll:false, limit:1, filters:{activeWorkflows:false,tags:"",name:""}. It sends /workflows?limit=1&active=false&name=: empty tags are omitted, empty name retained. Execution filters allow manual workflowId and status error/success/waiting. Workflow filters allow activeWorkflows, tags, name, projectId and excludePinnedData. Audit additionalOptions permits distinct categories credentials/database/nodes/filesystem/instance and optional daysAbandonedWorkflow 0–365000; it is a POST with provider-controlled effects, and HTTP 200 is not security certification.

Useful mappings include node.<id>.json.data.id, node.<id>.json.data.0.id, node.<id>.json.data.properties.region.type and node.<id>.json.data.reports.0.risk. Workflow parameters/connections/settings/staticData/pinnedData/credential payloads, arbitrary schema defaults/examples/const values and audit settings/version-node bodies are omitted under this explicitly narrower migration. Unknown schema keywords and malformed or over-budget shapes fail closed. paging.nextCursor authorizes no fetch: prepare and confirm a separate action. Execution new/running/waiting remains provider state and never starts polling. These examples are synthetic.

The original run actor must confirm the exact resolved request with current workspace AND organization execution access. Selected encrypted account revision, published node/version, run input, live-policy generation, original task ID AND wait token and execution lease remain bound through one winning durable claim, settlement and fresh authenticated output. A replaced task never renews consent. One physical attempt is allowed, including workflow maxTries3; timeout, reset, rate limit or an uncertain used claim never triggers automatic resend, readback, redirect or refresh.

Names, IDs, tags, schema option labels, job/report fields and audit locations may contain personal data retained in ordinary run, checkpoint, Human Task history and downstream mappings or AI. The key uses existing credential encryption; this adds no raw-database encryption, automatic deletion/TTL, scanner or deletion guarantee. Local bounds are 64 KiB input, 1 MiB complete response, depth 32, 20,000 nodes, 4096 reads and thirty seconds for the whole invocation, rather than vendor guarantees. Real provider authentication, selected-instance permissions/schema/version/entitlement, durable DB/Redis recovery, native DNS/TLS/sockets, Review tutorials and real AI request interpretation remain untested; local production-owner tests use simulated dependencies.

n8n destructive Public API actions

The selected n8n Public API subset now has nine ordinary actions. Delete a credential (credential.delete, n8n_public_api_credential_delete), Delete an execution (execution.delete, n8n_public_api_execution_delete) and Delete a workflow (workflow.delete, n8n_public_api_workflow_delete) each send one destructive WRITE: DELETE /credentials/<credentialId>, /executions/<executionId> or /workflows/<workflowId>, prefixed by the saved API root. There is no query or body. Credential/workflow IDs are explicit 1–256 ASCII letters, digits, underscore or hyphen; execution IDs are positive safe decimal strings. No collection, wildcard or returned URL becomes a deletion target.

The existing encrypted manual API-key profile and unverified instance edition/version/scopes/entitlement/free-trial prerequisites still apply. All three require the initially unset schema-2 migration, raw-retention and one-attempt controls plus destructiveAcknowledged:true, then a separate original-actor Human Task showing the saved account, origin/root, exact DELETE path and named resource ID. Imported string "true", 1, null and invalid objects remain visible and unchecked until explicitly repaired; opening or saving another field never grants consent. Current workspace and organization authority, policy/account revision, original taskId and waitTokenId, immutable request and lease/claim/CAS checks still apply, including later authenticated OUTPUT.

A genuinely empty HTTP204 or HTTP200 response maps to data:{resource,id,response:"empty"}, with zero examined/returned records and outcomeVerified:false; it observes a response and does not prove deletion. HTTP200 expected metadata maps to response:"metadata" and bounded metadata, with an exactly matching target ID. Credential metadata is only id/name/type; execution data omits runData/workflowData/customData; workflow data omits parameters/settings/pinnedData/connections/credential references. JSON null, whitespace, malformed/unexpected bodies, wrong IDs and other statuses are withheld. There is no deleted:true, automatic readback, paging, polling, retry or uncertain-effect replay. Map node.<id>.json.data.id, .response or available .metadata.name/.status; names and IDs may contain PII retained in ordinary history and downstream mappings, without a new raw-database encryption, TTL/deletion or scanner guarantee.

Real deletion effects, provider authentication, durable DB/Redis, native DNS/TLS and matching-revision Review remain untested. Use only separately authorized disposable resources for live verification; existing credentials, workflows and executions are not test targets. Credential create, workflow create/update/activate/deactivate/getVersion and Tool/Trigger remain unavailable; nine actions do not establish full-node parity.

A completed synthetic delete input, shown with explicitly chosen acknowledgments rather than defaults:

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "oneAttemptAcknowledged": true,
  "destructiveAcknowledged": true,
  "credentialId": "cred_A"
}

For execution deletion replace credentialId with executionId:"42"; for workflow deletion use workflowId:"wfA". Enter the ID and select the exact saved account manually. This example does not approve the separate Human Task or verify a provider effect.

AWS Transcribe job requests and metadata

The partial AWS Transcribe subset supports four ordinary numeric-version-1 actions: transcribe.transcriptionJob.create, transcribe.transcriptionJob.delete, transcribe.transcriptionJob.get and transcribe.transcriptionJob.getAll. Transcript downloading, simplification, automatic language detection, unbounded pagination and Tool/Trigger forms are unsupported. This is partial service coverage, not full-node or provider parity.

Save an existing awsTranscribeManualSigV4V1 account with profile aws-transcribe-manual-sigv4-v1, partition aws, one of twenty pinned standard regions, declared twelve-digit account ID and manual access keys. Named fields and Advanced JSON use the same encrypted, write-only account. manual-static omits token and expiry; manual-temporary requires an explicitly supplied session token and actual UTC expiry, with more than ninety seconds remaining at execution. Save/update/reopen is local and unverified with zero provider HTTP/DNS. Test and bulk Test are unavailable. No ambient/system/role credentials, STS acquisition, refresh, discovery or custom endpoints are used. The declared account does not prove AWS identity.

Import exact n8n-nodes-base.awsTranscribe with numeric typeVersion:1, preserve original parameters and credential references, and repair the schema-2 input deliberately. Select the exact saved account and set the initially unset migrationAcknowledged:true and rawRetentionAcknowledged:true controls. Imported strings such as "true", numbers and null remain invalid, unchecked and preserved until an explicit click or repair. Unknown and inactive fields remain repair items; embedded credentials are refused before source-ledger persistence. Original get defaults to returnTranscript:true; that intent is unsupported and is never silently converted to metadata.

Create requires an explicit unique transcriptionJobName, authorized same-region s3://bucket/key media URI, mediaFormat, detectLanguage:false, supported languageCode and an explicit options object. The original v1 node has no media-format field, so imports require manual format selection. Supported original settings include channel identification, maximum alternatives/speaker labels (2–10), vocabulary name and a complete vocabulary-filter name/method pair. Channel identification cannot combine with speaker labels. Deliberately set the initially unset processingAcknowledged:true: AWS obtains the authorized media, processes speech, may charge costs and retains job/transcript data under its policies; the app never fetches the media URI. A QUEUED or IN_PROGRESS result remains pending. No polling or completed-transcription claim follows submission.

Delete requires the explicit job name and initially unset destructiveAcknowledged:true, then a separate original-actor Human Task showing the exact AWS request. An actual empty HTTP200 body yields empty operationData:{} with responseReceived:true and outcomeVerified:false. It is response evidence only; job deletion completion and transcript object removal are not verified. There is no fabricated success:true, hidden readback or automatic resend.

Get requires returnTranscript:false explicitly. It returns bounded typed job/status/time/media/transcript-location metadata in operationData. Transcript locations remain inert JSON strings, including on COMPLETED jobs: the app never fetches a provider-returned URI or simplifies a transcript. Get All requires explicit returnAll:false, limit 1–100, filters:{} or supported original status/name filters, and boundedPageAcknowledged:true. It returns one partial page envelope with TranscriptionJobSummaries and any actual top-level NextToken. Copy that token into nextToken only for a separate manually approved next request. A short or empty page does not prove exhaustive inventory; the original broken unbounded paginator is not reproduced.

Only the original run actor can approve the exact request within fifteen minutes. Current workspace and organization authority, selected account, published node/version, run input, resolved request, original task ID, wait token, stable durable decision and execution lease are checked through claim, dispatch, settlement and fresh OUTPUT. Replacement never renews consent. One claim permits at most one physical attempt despite workflow retries; timeout, revocation, disconnect, expired lease or uncertain settlement cannot authorize resend. Same-lineage failure cleanup grants no new authority and remains fenced by the real lease.

Local limits are 16 KiB request, 1 MiB whole response, thirty seconds, depth32/nodes20000 and strict unknown-field refusal. Malformed, oversized, unexpected, numerically mistyped or credential-reflecting responses fail closed. Useful job metadata, names and media/transcript locations enter ordinary workflow/checkpoint/Human history and may reach downstream AI. Accepted raw response bytes use the existing encrypted protected Workspace Files mechanism; this adds no raw-database encryption, automatic TTL, plaintext-scanner or deletion guarantee.

En español: selecciona una cuenta manual existente y la acción exacta. Corrige los campos importados y marca cada confirmación deliberadamente. AWS obtiene el archivo autorizado; la aplicación no descarga archivos ni transcripciones. Los trabajos pendientes siguen pendientes, eliminar solo acredita la respuesta y listar devuelve una página parcial con un token manual. Los metadatos y sus ubicaciones pueden conservarse en el historial y llegar a la IA posterior.

Local tests simulate provider/Prisma facts and use actual public resume, Human Task, checkpoint and output owners. Real AWS authentication/permissions/effects, native DNS/TLS, DB/Redis restart, real AI request interpretation and Review workspace/organization Help walkthroughs remain untested because those external environments and credentials are outside this worker's authority. See the official StartTranscriptionJob, DeleteTranscriptionJob, GetTranscriptionJob and ListTranscriptionJobs contracts.

A manually authored create request after reviewing each acknowledgment:

{
  "transcriptionJobName": "manual-a",
  "mediaFileUri": "s3://authorized-input/a.wav",
  "mediaFormat": "wav",
  "detectLanguage": false,
  "languageCode": "en-US",
  "options": {},
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "processingAcknowledged": true
}

A separate one-page metadata list request:

{
  "returnAll": false,
  "limit": 1,
  "filters": {},
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "boundedPageAcknowledged": true
}

AWS Certificate Manager certificates

The five ordinary numeric-version-1 actions are acm.certificate.delete, acm.certificate.get, acm.certificate.getMany, acm.certificate.getMetadata and acm.certificate.renew. Select AWS Certificate Manager, then the exact certificate action. Tool/Trigger variants, certificate requests/import/export, private-key export and ACME management are unavailable.

Use an encrypted saved awsAcmManualSigV4V1 account with profile aws-acm-manual-sigv4-v1, partition aws, a supported pinned commercial region, declared twelve-digit account ID, access key ID and secret access key. Named fields and Advanced JSON save the same account. manual-static omits session token/expiry; manual-temporary requires the actual supplied session token and UTC expiry; execution requires more than ninety seconds remaining. Local Save can retain an expired profile as unverified, but execution refuses it without refresh. The current bounded endpoint profile supports the same twelve explicitly pinned regions as Comprehend; other regions and partitions require a separate profile. The declared account ID restricts the certificate ARN and is not verified identity. Save/update/reopen is local, encrypted, write-only and unverified, with zero provider HTTP/DNS. Test and bulk Test are unavailable. No automatic STS, AssumeRole, system credentials, refresh, custom/FIPS/dualstack endpoints or discovery occurs.

Import n8n-nodes-base.awsCertificateManager with numeric typeVersion:1, preserve its original node/parameters/credential references, then reselect the native account and repair schema 2 explicitly. Unknown and inactive values stay visible and refuse execution. Embedded credential material is refused before source-ledger/workflow persistence. Original authentication:iam maps to the manual account lifecycle; AssumeRole/system/custom endpoint settings require manual repair. Remove the original unused S3 bucketName and certificateKey fields explicitly; they never become ACM request fields. The original default renew is not consent. Portable imports and exports can retain non-secret domain/certificate business data and original references.

Set parameterSchemaVersion:2, migrationAcknowledged:true and rawRetentionAcknowledged:true. Acknowledgments start unset, require literal boolean true and retain imported string "true" as invalid and unchecked until explicitly repaired. Delete also requires destructiveAcknowledged:true; renew requires providerEffectsAcknowledged:true. The exact certificate ARN must match the selected region/partition/account. Only the original run actor can approve the exact named request within fifteen minutes, with current workspace AND organization authority. The runtime captures the original task ID, wait token and durable decision at resume production, then loads that original task by ID throughout claim, send, settlement and fresh OUTPUT. A replacement task, wait, decision, account, node or input cannot renew consent. The execution lease remains required; an expired lease cannot weaken the physical-claim or cleanup fence.

Delete removes the certificate and associated private key; an in-use certificate cannot be deleted. AWS documents eventual consistency and unchanged Private CA billing. Renew requests renewal of an eligible certificate and can require Private CA permissions and incur costs. Both are writes: an empty HTTP 200 proves only that the response was observed, and the result retains outcomeVerified:false. There is no completed-deletion/renewal claim, readback, polling or certificate acquisition. Get returns only valid certificate PEM blocks and optional certificate chain. Get Metadata returns useful bounded exact certificate/domain/subject/contact/status/time/algorithm/type data. Those fields remain available for downstream mapping.

Get Many reads exactly one page, with original limit 1–500/default100, returnAll:false and boundedPageAcknowledged:true. Preserve the nested original options.certificateStatuses, options.extendedKeyUsage, options.keyTypes and options.keyUsage arrays. Extended Key Usage applies independently of status, correcting the original gate. Omit unused filters; explicit empty arrays require repair. AWS defaults restrict types to RSA_2048 and omit ACME origins, so a short/empty page does not prove exhaustive account inventory. NextToken remains returned business data and is not followed. Return All, continuation input and newer sort/origin/export filters require separate explicit extensions.

Local bounds are 16 KiB request, 1 MiB whole response, thirty seconds, depth32/nodes20000 and one physical attempt, even when workflow retries are configured. AWS permits a GetCertificate chain up to 2 MiB; this application's narrower whole-response bound refuses a valid oversized chain without truncation or fabricated success. Malformed/duplicate/invalid JSON, unexpected fields/status, credential reflections, private/reserved origins and partial responses fail closed. Once claimed, a timeout, disconnect, parser/storage failure or uncertain outcome never permits replay through a new gateway.

Exact accepted business output, inputs, domain names, ARNs, subjects, validation contacts and certificate PEM may remain in ordinary run/checkpoint/Human history and reach downstream AI or remote actions. Accepted raw response bytes are application-encrypted into protected Workspace Files and authenticated at fresh OUTPUT; the configured scanner sees ciphertext. This supplies no raw database encryption, automatic TTL, plaintext scanner or deletion-completion guarantee. Invalid responses and credential/error reflections do not become accepted business output.

For example, one partial list uses:

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "returnAll": false,
  "limit": 100,
  "boundedPageAcknowledged": true,
  "options": { "extendedKeyUsage": ["TLS_WEB_SERVER_AUTHENTICATION"], "keyTypes": ["RSA_2048"] }
}

An explicitly acknowledged deletion names its certificate:

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "certificateArn": "arn:aws:acm:us-east-1:123456789012:certificate/12345678-1234-1234-1234-123456789012",
  "destructiveAcknowledged": true
}

En español: elige la acción exacta y una cuenta manual guardada, corrige los campos importados y marca cada confirmación deliberadamente. El actor original confirma el ARN y la solicitud; HTTP 200 solo demuestra una respuesta, no la eliminación ni la renovación finalizadas. Una lista lee una página parcial y los datos del certificado permanecen disponibles para pasos posteriores con la retención habitual.

Local tests use simulated provider/store boundaries. Real AWS permissions/eligibility/effects, native DNS/TLS/provider execution, durable DB/Redis restart, storage scanning and Review workspace/organization Help walkthroughs remain untested. See the official DeleteCertificate, GetCertificate, ListCertificates, DescribeCertificate and RenewCertificate contracts.

DynamoDB typed items

Select AWS DynamoDB for the four ordinary numeric-version-1 operations dynamodb.item.upsert, dynamodb.item.delete, dynamodb.item.get and dynamodb.item.getAll. These use PutItem, DeleteItem, GetItem and an explicit Query or Scan. This is a partial service integration; automatic mapping, simplified decoding, discovery, Tool/Trigger variants and other APIs are unavailable.

Save awsDynamoDbManualSigV4V1 with profile aws-dynamodb-manual-sigv4-v1, partition aws, one of twelve pinned commercial regions, a declared twelve-digit account, and manual static keys or the supplied temporary session token with its actual UTC expiry. Temporary execution requires more than ninety seconds remaining. Named fields and Advanced JSON save/update/reopen the same encrypted write-only, locally unverified account with zero provider HTTP/DNS. Test and bulk Test are unavailable; no SDK, environment/metadata discovery, role assumption, STS acquisition, refresh or custom/FIPS/dualstack endpoint runs. A declared account is not verified AWS identity.

Import exact n8n-nodes-base.awsDynamoDb with numeric typeVersion:1. Original parameters, raw JSON types and credential references remain retained through edit, serialize, reopen and reimport. Reselect the typed saved account and deliberately repair parameterSchemaVersion:2, typed item/key/expression fields and unsupported legacy options. Embedded credential controls are refused before source-ledger persistence; ordinary typed database attribute names remain business data. The initially unset migrationAcknowledged, rawRetentionAcknowledged and onePhysicalAttemptAcknowledged must each be literal boolean true. Imported "true", 1 or null is never consent.

PutItem is full replacement, including removal of omitted attributes on an existing item with the same key. Supply the complete typed item and explicit replaceAcknowledged:true. Delete requires an exact typed key and deleteAcknowledged:true; choose returnValues:NONE or ALL_OLD. Both also require a separate original-actor Human Task for the exact table/item or key, conditions, account and consequences. Conditions can fail without retry or hidden readback. acknowledged:true, submitted input and returned old attributes observe the response; remotePersistenceVerified:false and outer outcomeVerified:false remain explicit. They never prove remote persistence.

AttributeValue JSON preserves S, exact decimal-string N (up to38 significant digits within the documented range), strict padded base64 B, BOOL, NULL:true, nonempty unique SS/NS/BS, L and M, including empty lists/maps. Primary keys use nonempty scalar S/N/B values. Attribute names and table schema are entered manually, without discovery. Set/list and data-name bounds may be narrower than AWS. GetItem allows explicit consistency and projection aliases; a missing Item returns found:false, while malformed or unexpected bodies fail closed.

Get All requires an explicit mode:Query with keyConditionExpression, or mode:Scan; index, filter, projection, aliases and typed values are manual. Set returnAll:false and limit:1..100: Limit counts evaluated items, not returned items or a cost guarantee. Exactly one bounded page is read. An empty filtered Items array can still return a typed lastEvaluatedKey; copy it to exclusiveStartKey only in a new, separately confirmed request. select:COUNT returns a count/scanned-count object, never an array. COUNT cannot include a projection. pageComplete means this response had no continuation key, not snapshot or exhaustive-inventory proof.

Local whole-input/request bounds are64KiB, whole response1MiB, depth32,20000nodes,4096reads and thirty seconds. Inherited32KiB signed approval and48KiB Human instructions can impose a smaller effective input. Oversized, missing, malformed, duplicate or unexpected bodies are refused without truncation. One durable physical claim remains bound to the actual public resume producer's original task ID, wait token and durable decision, current workspace and organization authority, account/input revision and live acquired lease through settlement and fresh authenticated OUTPUT. Expired ownership or an uncertain used claim cannot replay through a new gateway; unchanged-lineage failure cleanup uses the execution-lease CAS fence.

Table names, keys, attribute names and item data may be PII retained in ordinary workflow/checkpoint/Human history and downstream mapping or AI. Account encryption adds no raw-database encryption, automatic TTL/deletion or scanner guarantee. Local tests simulate provider/store facts; live AWS, native DNS/TLS, real DB/Redis restart, real AI phrasing and matching-revision Review workspace/organization Help walkthroughs remain untested. See official GetItem, DeleteItem and AttributeValue contracts.

Example typed read

After reviewing the retention and one-request decisions, configure this synthetic native input with your selected saved account:

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "onePhysicalAttemptAcknowledged": true,
  "tableName": "SyntheticRecords",
  "key": { "pk": { "S": "synthetic#1" } },
  "consistentRead": true,
  "projectionExpression": "#amount",
  "expressionAttributeNames": { "#amount": "amount" }
}

A returned amount such as {"N":"9007199254740993"} remains that exact string. A missing item instead returns operationData:{found:false}. Reads and writes can incur AWS charges; evaluated-item and byte limits are not a billing cap.

AWS ELB: seven ordinary actions

AWS ELB has partial service coverage for seven ordinary numeric-version-1 actions. Each sends one signed AWS Query GET, including writes, with Version=2015-12-01. Approval and permissions follow the actual AWS Action. Tool/Trigger variants and full node parity are unsupported.

Save awsElbManualSigV4V1 with profile aws-elb-manual-sigv4-v1, partition aws, one of the twelve pinned commercial regions, declared twelve-digit account ID, explicit manual-static keys or manual-temporary supplied token and actual UTC expiresAt. Temporary execution needs more than ninety seconds remaining. Named fields and Advanced JSON use the same encrypted write-only account; Save/update/reopen sends zero HTTP/DNS and stays unverified. No Test, bulk Test, ambient/system/role/metadata credentials, STS, token acquisition, refresh or discovery occurs. The fixed endpoint is https://elasticloadbalancing.<region>.amazonaws.com:443; the declared account is a restriction, not verified identity.

Import exactly n8n-nodes-base.awsElb with numeric typeVersion:1. Original node, parameters and credential references remain in awsSourceLedger; reselect the new saved account and manually set parameterSchemaVersion:2, migrationAcknowledged:true and rawRetentionAcknowledged:true. All acknowledgments start unset and require literal boolean true. Imported string "true", 1, null or objects remain visible and unchecked until clicked; portable serialize/reopen/reimport preserves that repair intent. Embedded credential controls are refused before source-ledger persistence. Unsupported and inactive values are never silently discarded.

For listener actions, supply exact listenerId (ListenerArn) manually. Add/remove also require one explicit certificateId (ACM ARN) and listenerChangeAcknowledged:true, all in the saved account and region. Optional original loadBalancerId is context only and must match the explicit listener. No DescribeListeners lookup runs. Adding changes the HTTPS/TLS certificate list; removal can break clients and cannot remove the default certificate. IAM certificate ARNs stay unsupported repair items; do not guess an ACM ARN or replace the default certificate here.

Create requires exact original name, schema (serialized as AWS Scheme), type (application or network), ipAddressType (ipv4 or dualstack) and manual subnets. Application creation in this subset requires at least two subnets in distinct Availability Zones, which the operator must verify; internal requires ipv4. Optional additionalFields.securityGroups and additionalFields.tagsUi.tagValues keep their original nesting. Local maxima are sixteen unique subnet IDs, five unique security-group IDs and fifty unique tag keys; tag keys/values obey AWS's character pattern and narrower local128/256 UTF-8-byte limits. Name is at most32 alphanumeric/hyphen characters, no leading/trailing hyphen or internal- prefix. Explicit provisioningAcknowledged, networkAccessAcknowledged and costAcknowledged authorize provisioning, ongoing costs and the chosen network exposure. No EC2, subnet or security-group loader runs. Gateway, newer IP modes, subnet mappings, Outposts/Local Zones and other extensions require manual repair.

Delete requires the exact loadBalancerId and destructiveAcknowledged:true: it also deletes listeners and can disrupt traffic. Deletion protection can reject it; registered targets continue running. Add/create return bounded exact certificate or load-balancer metadata, including ARN, name, DNS, state, type and subnet metadata when present. Get checks the returned ARN. Delete/remove accept an observed empty XML result and retain responseReceived:true, outcomeVerified:false; they never invent success:true, sucess:true or completed remote effects. A create response may still report provisioning.

Both Get Many actions require returnAll:false, limit1–400/default100 and boundedPageAcknowledged:true. They read one partial page. An exact opaque marker may be copied manually from an earlier NextMarker; no automatic continuation occurs. Original load-balancer filters.names is active only under unsupported Return All: supplied names, even inactive/empty source values, remain repair items and block execution until explicitly removed. There is no silent broader inventory query. Empty, singleton and repeated XML member lists preserve arrays and exact string data; IsDefault is a strict boolean. Duplicate/default certificate entries remain in order.

Only the original run actor may confirm the exact resolved account, action, input and request within fifteen minutes, with current workspace and organization access. Actual public resume preparation captures the immutable task ID, wait token and stable durable decision. They remain checked with account/input/node/policy revisions and the execution lease before claim, send, response, settlement and fresh OUTPUT. One actual checkpoint CAS permits at most one physical request, including configured workflow retries. Expired/uncertain claims never resend or adopt replacement consent; identity-only failed cleanup is fenced by the actual lease CAS.

Map useful data from node.<id>.json.operationData.LoadBalancers[0].DNSName, .LoadBalancerArn, .State.Code, .AvailabilityZones, or .Certificates[0].CertificateArn; operationData.NextMarker is manual continuation data. ARNs, DNS names, subnets, tags and raw XML can contain sensitive data retained in ordinary run/checkpoint/Human Task history and downstream mapping or AI. Accepted exact raw bytes use the existing protected Workspace Files encryption and fresh OUTPUT binding; this adds no raw-database encryption, automatic expiry or plaintext-scanner promise. Local32KiB request/1MiB response/thirty-second/depth32/nodes10000 bounds refuse oversize data without truncation. DTD/XXE/custom entities, malformed/duplicate XML, unexpected fields/status, credential reflection and invalid UTF-8 fail closed before accepted artifact retention. No redirects, returned-URL fetch, retries, polling or readback occur.

Local tests simulate provider/Prisma facts. Real AWS permissions, provisioning/TLS/deletion effects, native DNS/TLS/sockets, durable DB/Redis recovery and matching-revision Review workspace/organization Help walkthroughs remain untested. Official API contracts: CreateLoadBalancer, DeleteLoadBalancer, DescribeLoadBalancers, AddListenerCertificates, RemoveListenerCertificates and DescribeListenerCertificates.

AWS Cognito User Pools blocks

Partial ordinary numeric-version-1 actions: cognito.group.create, cognito.group.delete, cognito.group.get, cognito.group.getAll, cognito.group.update, cognito.user.addToGroup, cognito.user.create, cognito.user.delete, cognito.user.get, cognito.user.getAll, cognito.user.removeFromGroup, cognito.user.update, cognito.userPool.get. No OAuth/authentication/login, Tool/Trigger or full-node parity. Manual IDs replace dynamic loaders without identity guessing.

A one-page user list uses the exact PaginationToken; a group list instead uses nextToken/NextToken. Each later page needs independent approval.

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "userPoolId": "us-east-1_ExamplePool",
  "returnAll": false,
  "limit": 20,
  "boundedPageAcknowledged": true,
  "paginationToken": "opaque-token-from-prior-user-page"
}

A group update preserves zero precedence and explicitly acknowledges its write.

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "userPoolId": "us-east-1_ExamplePool",
  "groupName": "Research",
  "precedence": 0,
  "writeAcknowledged": true
}

Map useful metadata from operationData. UserAttributes/Attributes retain exact Name/Value arrays; they never overwrite Username or other metadata. Blank writes expose responseReceived only, never verified completion. See manual credentials, creation/message/alias consequences and response limits.

AWS IAM identity and membership actions

The ordinary numeric-v1 IAM user/group subset covers twelve actions only. It uses the dedicated awsIamManualSigV4V1 account: profile aws-iam-manual-sigv4-v1, partition aws, signing region us-east-1, and fixed global https://iam.amazonaws.com/ over HTTPS443. Use named fields or Advanced JSON for manual static keys, or a supplied temporary token with its actual UTC expiry. Save and reopen are encrypted, write-only, local and unverified; Test is disabled. No ambient, system, role, metadata, STS, credential acquisition or refresh is used.

Exactly one requested Action is sent by form POST with Version 2010-05-08. Identity, access and membership changes require explicit acknowledgments and a separate original-actor Human confirmation. DeleteUser/DeleteGroup fail on provider dependencies; they never remove memberships, login profiles, MFA, policies or access keys implicitly. Membership add/remove are separate exact user/group operations. CreateUser supports path, tags and permissions boundary; updates support path and rename only. No policy, role, access-key or login-password CRUD is added. HTTP200 is provider response evidence, never verified effect completion.

Lists require explicit returnAll: false, limit 1–1000 and bounded-page consent. GetGroup also pages Users; IsTruncated and opaque exact Marker remain visible even when Users are omitted. Continue only with another separately approved request. No automatic paging, membership scans, discovery, loaders, post-read, retries, returned-URL fetching or redirects. Original helper, returnAll and includeUsers semantics that cannot be preserved remain visible for manual repair; unknown fields and invalid consent types survive export/reopen. Embedded credentials are refused before source-ledger persistence.

Typed user/group identity, ARN, path, creation time, tags, boundary and explicitly selected membership data plus an exact raw reference enter ordinary run/checkpoint/Human history and may reach downstream AI. Account encryption adds no raw database encryption, automatic TTL or scanner guarantee. A bounded XML decoder rejects DTD, external/custom entities, duplicate or wrong envelopes, malformed members, unknown fields/status and credential reflection before retaining raw data. Current workspace and organization permissions, account, input, policy, immutable original task/wait/decision and lease are rechecked through claim, send, response, settlement and fresh OUTPUT. Uncertain or expired claims cannot resend. Local bounds: 32 KiB request, 1 MiB response, 30 seconds. Real AWS, hosted Review and full node/provider parity remain untested.

Supported explicit actions: iam.group.create, iam.group.delete, iam.group.get, iam.group.getAll, iam.group.update, iam.user.addToGroup, iam.user.create, iam.user.delete, iam.user.get, iam.user.getAll, iam.user.removeFromGroup, iam.user.update.

Example CreateUser input (review before explicitly granting each consent):

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "identityChangeAcknowledged": true,
  "userName": "Alice",
  "additionalFields": {}
}

AWS S3 ordinary bucket subset

The partial ordinary S3 subset covers only s3.bucket.create, s3.bucket.delete, s3.bucket.getAll and s3.bucket.search: four of twelve actions, with exact numeric source V1/V2 imports analyzed separately. Search is ListObjectsV2 object metadata, not a bucket-name search. Object upload/download/copy/delete, folder deletion, multipart, Tool and Trigger remain unsupported.

Save awsS3BucketManualSigV4V1 / aws-s3-manual-sigv4-v1 with partition aws, an explicitly known us-east-1 or eu-west-1 region, declared twelve-digit account ID and manual static keys or supplied temporary token plus actual UTC expiry. Temporary execution needs over ninety seconds remaining. Named fields and Advanced JSON use encrypted, write-only, unverified local Save/update/reopen with zero provider HTTP/DNS. Test, bulk Test, STS, system/role chains, refresh, custom endpoints and discovery are unavailable.

Every request requires schema2 and initially unset migrationAcknowledged, rawRetentionAcknowledged, regionAcknowledged and billingAcknowledged, all literal true, plus action-specific consent. Raw string "true", 1, null and objects remain visible and unchecked until a deliberate click. Then a separate original-actor Human Task approves the exact request. Current workspace and organization access, account, node, input, policy, original task ID/wait token/stable decision and real execution lease are checked through the one durable claim, send, response, settlement and fresh OUTPUT. Expired or uncertain claims never replay.

Create requires name, additionalFields:{} (or a matching known region) and name/global-namespace/security/write acknowledgments. It sends one PUT: empty body for us-east-1, matching XML LocationConstraint elsewhere. Only provider secure defaults are supported; all original ACL/grant/ownership/tag/object-lock intent stays for explicit repair. HTTP200 empty body yields response evidence with effectConfirmed:false; Location remains inert manual metadata. Creating an existing owned us-east-1 bucket can reset ACLs under AWS behavior; inspect the exact name and permissions manually before approval.

Delete requires name, empty-bucket/global-namespace/destructive acknowledgments and sends one DELETE. The bucket must already be empty of objects, versions and delete markers; nothing enumerates or deletes them or changes protection/ownership. Only HTTP204 empty body is accepted as response evidence, never confirmed deletion. A deleted global name may be reused by another account.

ListBuckets (getAll) needs explicit returnAll:false, limit:1..1000, filters:{} and boundedPageAcknowledged:true. This is a local cap, not the AWS maximum10000. Optional filters.bucketRegion must equal the endpoint region; filters.prefix filters bucket names. Output preserves actual ContinuationToken; copy it byte-for-byte into nextToken for a separately approved request. Unpaginated accounts with approved quota above10000 may be rejected; no auto-page or local slice is used.

ListObjectsV2 (search) needs bucketName, explicit returnAll:false, limit:1..1000, additionalFields:{}, bounded-page and Requester Pays acknowledgment. Supported typed options are prefix, delimiter, encodingType:url, fetchOwner, startAfter and requesterPays; the payer is a header. Dotted names use regional HTTPS443 path style. Output keeps bounded Contents, CommonPrefixes, KeyCount, IsTruncated, keys, ETags, sizes, storage classes, dates and actual NextContinuationToken; tokens and keys remain inert values. No content or returned URL is fetched. Empty XML containers and singleton/many entries normalize without fabricated rows.

Original versions keep exact parameters and credential references. V1 ignores the location region during signing; V2 forwards it and handles dotted search names differently. Both original location reads, null/us-east-1/EU assumptions, returnAll/default/limit and unsupported options require visible manual repair. No hidden GetBucketLocation/HeadBucket/readback, redirect, follow-up, polling or retry occurs. Local32KiB request/1MiB response/thirty-second bounds apply; malformed, duplicate, DTD/entity/XXE, wrong-envelope, oversized or credential-reflecting replies fail before accepted artifact retention. Useful metadata and tokens enter ordinary run/Human history and downstream AI; account encryption adds no raw DB encryption, TTL or scanner guarantee.

En español: selecciona una cuenta manual existente y una región conocida, corrige el origen V1/V2 y marca cada confirmación de forma deliberada. Crear y eliminar solo acreditan respuestas; listar devuelve una página parcial y buscar lista metadatos de objetos. Copia el token exacto para otra solicitud aprobada. No se descargan objetos ni se eliminan sus versiones.

Local tests simulate provider/Prisma facts. Real AWS credentials, permissions/effects, native DNS/TLS, DB/Redis restart, AI request interpretation and Review workspace/organization Help walkthroughs remain untested because they are outside this worker's authority. See CreateBucket, DeleteBucket, ListBuckets and ListObjectsV2.

Venafi certificate reads

The two public preview routes are typed, separate services:

Set parameterSchemaVersion: 2, migrationAcknowledged, rawRetentionAcknowledged and onePhysicalAttemptAcknowledged explicitly. Consent starts unset; invalid imported strings, numbers, objects or null remain visible and unchecked until repaired. Other actions, options, filters, binary data and async flows require manual repair. Select the matching saved account and confirm the exact request in its separate original-actor Human Task.

Cloud returns useful certificate IDs, subject names, validity and unknown bounded business fields. Its typed page reports returned, hasNext (null when unknown), completeness and automaticFollow: false. A Next value is inert data: it is never fetched. Empty, partial and unknown pages remain distinct; one page never proves the whole inventory was retrieved. The current local unit accepts only limit 1; larger limits and traversal require a separate reviewed extension.

Datacenter requires the returned Guid to match the submitted representation exactly and preserves that source ID. Use brace-enclosed GUIDs as documented by Datacenter 26.1; other representations remain unverified. It preserves optional Name, DN, CertificateDetails, ValidationDetails and unknown bounded fields. ValidationState: "Failure" is retrieved business state, not successful validation; missing validation details do not certify validation either. HTTP 404/no record, authentication/permission/rate/transport errors, malformed envelopes, duplicate JSON keys, unsafe numbers, invalid UTF8 and oversize replies cannot become successful data or retained raw artifacts.

Accepted raw UTF8 is authenticated-encrypted before checkpoint/history retention and bound to the original scope, task, wait and stable decision. Useful projected business fields and source items remain in ordinary authorized workflow history and may reach downstream AI; certificate inventories can reveal internal hostnames and personal data. This adds no database/backup encryption, automatic deletion, TTL or plaintext scanning guarantee. No real-provider, native socket, persistent DB/Redis restart or matching-revision Review verification has been performed by the local simulated tests.

Local bounds are 64 KiB input/request, 1 MiB response, depth 32, 20,000 JSON nodes, 4,096 stream reads, a 30-second request deadline and one physical attempt. Failed, expired or ambiguous attempts never automatically retry, poll, page, redirect, follow a returned URL or read back the certificate. A retrieved certificate does not prove its validity or successful provisioning.

TheHive v1 manual case reads

TheHive and TheHiveProject are two imported wrappers of one Query API service. Only exact ordinary version-1 n8n-nodes-base.theHive and n8n-nodes-base.theHiveProject with explicit case/get are supported. Choose case.getLegacy with theHiveApi and explicit apiVersion: "v1", or case.getProject with theHiveProjectApi. Legacy input is { "id": "~case-id" }; Project input is { "caseId": "~case-id" }. Project also accepts the original manual ID locator. Tool/Trigger variants, missing versions, v0, writes, options, arbitrary queries, list/URL locators, loaders, binary and asynchronous actions require manual repair. Original source controls remain retained and checked.

Save an existing API key with an explicit public HTTPS origin (no subpath, credentials, query, fragment or custom port), exact wrapper profile and an explicit organization policy:

{
  "profile": "theHiveApi",
  "apiVersion": "v1",
  "url": "https://incident.example.org",
  "ApiKey": "SYNTHETIC_EXISTING_API_KEY",
  "organization": { "mode": "named", "name": "responders" },
  "allowUnauthorizedCerts": false
}

For Project, use profile: "theHiveProjectApi" and omit apiVersion. Named fields and Advanced JSON use the same validator; save, update and reopen are local and unverified, with zero provider HTTP. Test and bulk testing are unavailable. Choose { "mode": "default" } explicitly to use the API key owner's default organization, or named scope to send X-Organisation. The application does not discover or verify that organization. A private-only installation is outside this public-origin policy; deployment outbound policy and reachability remain unverified.

Both reads need a separate signed 15-minute Human Task decided by the original run actor. Current workspace and organization membership, account, workflow, input, policy, task, wait, durable decision and execution lease must still match. One physical claim permits one bounded POST to /api/v1/query?name=get-case-…; retries, polling, paging, refresh and returned-link fetching are unavailable even if node retries are enabled. Legacy sends only getCase; Project adds exactly page 0 to 10 with extraData: ["attachmentCount"].

Outputs retain the wrapper's sourceIdentity, case ID, case number, title, unknown business fields and Project attachment count. Zero results return kind: "no-record"; incomplete/malformed/error/mismatched responses are refused. Numeric tokens that cannot pass the explicit precision policy are refused instead of rounded. Only a fully validated bounded response may be retained as an application-encrypted Workspace File; the configured scanner sees ciphertext. Useful business data can persist in workflow history and downstream steps. Fresh OUTPUT admission verifies current authority and the exact signed receipt and raw artifact, with no resend after uncertainty. Installed API version, provider acceptance, key permissions, actual organization visibility, native network behavior and matching-revision Review remain unverified; this is partial support.

TheHive case read field reference

thehive case.getLegacy

thehive case.getProject

Odoo contact reads

contact.get requires a canonical positive safe decimal-text contactId and options.fieldsList with1–32 unique names. contact.getAll requires actual returnAll:false, numeric offset0–1000000, numeric limit1–100 and selected fields. A missing/unknown filter, custom model, write, loader, binary/async or unsupported option requires visible manual repair. Business strings and unknown response fields are preserved; unsafe numeric precision refuses. One total30s budget spans both requests, with no retry or automatic paging.

Deployment/version, Custom plan, model, ACL and public reachability are unverified. Local provider/Prisma facts are simulated; real provider/native/DB/Review organization and workspace behavior is untested. See the complete bounded protocol and limitations.

Flow task/get partial compatibility

Only ordinary n8n-nodes-base.flow numeric typeVersion1 task/get is supported. This is partial compatibility derived from the shipped node and helper, with live protocol and account entitlement unverified. A bounded unauthenticated GET of the official developer documentation failed DNS ([Errno 8] nodename nor servname provided, or not known); that failure does not show provider retirement.

Save flowManualBearerV1 with named fields or Advanced JSON: {"profile":"flowManualBearerV1","organizationId":"701","accessToken":"YOUR_EXISTING_TOKEN"}. Enter secrets only in the encrypted account editor. Save/update/reopen is local, write-only and unverified with zero provider HTTP; Test is unsupported and bulk Test skips it. No OAuth, token discovery, acquisition or refresh occurs. Provider organizationId is independent of application organization/workspace IDs and never inferred from them.

Use task.get with parameterSchemaVersion:2, migrationAcknowledged:true, rawRetentionAcknowledged:true, onePhysicalAttemptAcknowledged:true and one exact positive decimal taskId. A safe integer input is accepted; unsafe numbers refuse rather than round. Decimal strings preserve large IDs. Optional filters.include accepts only schedule, files, file_associations, parent, at most once each. Example: {"parameterSchemaVersion":2,"migrationAcknowledged":true,"rawRetentionAcknowledged":true,"onePhysicalAttemptAcknowledged":true,"taskId":"9007199254740993123","filters":{"include":["files"]}}. Acknowledgments start unset and require explicit review. Unknown optional controls remain visible manual repair intent; imported source parameters, versions and credential references remain in a source ledger. Reselect the typed saved account deliberately.

The original actor with current workspace AND application organization authority separately approves exactly one GET https://api.getflow.com/v2/tasks/{taskId}?organization_id={credentialOrganizationId} with Bearer accessToken. Original task ID, wait token, stable decision revision, actor/account/input/node/policy and execution lease remain bound through one physical claim CAS, response, settlement and fresh OUTPUT. Ambiguous responses, changed ownership or expired leases never grant a resend, including with node Retry enabled. Local limits are thirty seconds, 1MiB response, depth32 and 20k nodes; these are not provider limits. Self-hosted outbound reachability remains unverified.

Use data.task and optional included entities from the intact response envelope. Unknown useful business fields remain available; every provider JSON number is represented as its exact original text, including precision and exponent spelling. Original source IDs are not replaced by labels or application IDs. Accepted raw UTF-8 is application-encrypted in scoped WorkspaceFiles and represented by rawReference; plaintext raw/sourceItems are absent from ordinary output. Business data can remain in workflow history and downstream mappings/AI. The scanner sees ciphertext, not plaintext; no raw-database encryption, deletion deadline or backup deletion claim is made. Invalid envelopes, business errors, missing/mismatched tasks, auth/status/transport/rate failures, duplicate JSON, oversize bodies, invalid UTF-8 and secret reflection refuse success and accepted raw retention.

Create/update/getAll, triggers, webhooks, dynamic loaders, paging, Link helpers, retries, polling, readback and returned-URL fetch are unsupported. Mocked provider and Prisma facts do not verify live Flow, native DNS/TLS, durable DB/Redis, real AI or Review workspace/organization walkthroughs.

Splunk existing search job metadata

Exact ordinary n8n-nodes-base.splunk, numeric typeVersion:2, supports one action: Get existing search job metadata, search.get, tool splunk_search_get. Enter an existing searchJobId of 1–256 ASCII letters/digits/underscore/hyphen/dot, beginning with a letter, digit, underscore or hyphen. Named and JSON fields use the same strict parser. Set parameterSchemaVersion:2 and deliberately select the initially unchecked migrationAcknowledged, rawRetentionAcknowledged and onePhysicalAttemptAcknowledged controls. Importing never grants consent.

{
  "parameterSchemaVersion": 2,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "onePhysicalAttemptAcknowledged": true,
  "searchJobId": "1700000000.17"
}

The request is exactly GET /services/search/jobs/1700000000.17?output_mode=json, with the selected original management Bearer token. It does not create a job or retrieve results. The bounded response must contain exactly one matching entry with useful dispatchState metadata. Empty entries, partial/list envelopes, business error messages, malformed JSON and unsafe numeric JSON cannot become success. Content fields override outer entry fields; outer link is omitted. The effective URL-shaped id becomes entryUrl, then its last path segment becomes id. entryUrl and all other returned URLs remain inert data. Numeric strings such as "900719925474099312345" stay unchanged, and unknown business fields remain available.

Use result.data.id, .dispatchState or actual numeric-string count fields in downstream mapping. sourceItemsJson preserves the corresponding formatted item. Useful metadata remains ordinary authorized history/downstream data, while the original bounded response is retained as an authenticated encrypted protectedPayload bound to account and action; this adds no database encryption, TTL, scanner or deletion guarantee. outcomeVerified:false and completeness unknown remain explicit: metadata, even DONE, does not prove completed-job or result completeness. Local bounds are 16KiB input, 1MiB response, depth32, 20,000 nodes, 4096 reads and one 30-second invocation. No retry, paging, polling, redirect, returned-URL fetch, v1, HEC, create/search/SPL/run/results/report/alert/user parity is claimed.

Azure Cosmos DB: get one existing container

The three acknowledgment switches start unchecked on a new or original n8n import. Choose each explicitly. Only literal boolean true counts as consent; imported strings such as "true", other invalid JSON values and expressions remain invalid through unrelated edits and export/reopen until explicitly repaired.

Select Azure Cosmos DB → Get container for the ordinary numeric n8n version 1 container/get subset. Save an existing account in Named fields or Advanced: edit JSON using exactly:

{
  "profile": "azureCosmosDbMasterKeyV1",
  "account": "your-public-account",
  "database": "ExactDatabaseId",
  "key": "YOUR_EXISTING_CANONICAL_BASE64_MASTER_KEY"
}

Account is the public Azure account name, without a URL. Database and container IDs preserve case and leading zeros; the initial supported ID alphabet is ASCII letters, digits, underscore, plus, equals, comma, dot, at sign and hyphen, 1–255 characters, excluding . and ... Each path segment is encoded once. The database belongs to the credential; the action supplies container. Prefer an existing read-only master key. Save/update/reopen encrypt locally, send zero HTTP and remain unverified; Test is unavailable and bulk Test skips the account. SQL/NoSQL API, local key auth, read-only-key rights, RU charges and network ACLs remain unverified. No Azure resource or identity is created.

Reselect the exact saved account and explicitly check schema-2 migration, normal-history retention and one-attempt acknowledgments. Enable WORKFLOW_LIVE_AZURE_COSMOS_DB=true independently of the current workspace Azure Cosmos DB live preference. The original actor needs current direct organization and workspace membership. One separate Human Task binds actor, organization, workflow/version/node, account ID/revision, origin, database, container and retention. Name matching or another account never substitutes for a deleted, rotated or unauthorized selected account. Approval expiry, cancellation, lease loss or any changed binding refuses the request.

The runtime makes one HMAC-SHA256 authenticated HTTPS GET to https://{account}.documents.azure.com/dbs/{database}/colls/{container}, with x-ms-date freshly generated at dispatch and x-ms-version: 2018-12-31. It signs the lower-case verb, resource type and date plus the case-sensitive encoded resource path, including the final blank line. Keys are decoded from canonical base64; no unsigned fallback exists. Public DNS and TLS are pinned and revalidated; private/custom/sovereign origins, alternate ports and redirects are refused. No item or partition-key probe, query, pagination, retries, management API or returned-link fetch occurs.

A successful 200 returns data.container, preserving exact string IDs, resource metadata and bounded unknown JSON business fields, including nested fields called token, password, __proto__ or constructor. Those values can contain private data and remain in ordinary history and downstream mappings. Raw transport text is discarded. Actual credential value echoes, duplicate JSON keys, unsafe integers, overflow, malformed UTF-8/JSON, mismatched ID, empty body and status errors refuse success. Limits are one total 30-second dispatch/read/decode budget, 1 MiB response, depth 32, 20,000 nodes and 4,096 reader steps. A durable claim and consumed-decision annotation are written together before HTTP; uncertainty or removed/tampered effects never authorize replay.

Original imports require exact ordinary numeric version 1, explicit container/get, By ID locator and simple:false. Original omitted container operation defaults to getAll and omitted simplify defaults true; unsupported controls outside the admitted listing profile require repair; source controls outside this original read contract, Tool/Trigger, extra auth/parameters and dynamic locators require visible manual repair. Source intent remains in the import ledger. OAuth, managed identity and connection strings are unsupported. Local simulations do not verify real Azure, native deployment, durable database/Redis restart or Review organization/tutorial behavior.

YouTube one video snippet

Only ordinary n8n-nodes-base.youTube, numeric typeVersion:1, resource:"video", operation:"get", and explicitly selected part:["snippet"] are supported. The native operation is video.get, tool youtube_video_get. Enter one literal video ID of 11 ASCII letters, digits, underscores or hyphens; URLs, multiple IDs, numeric IDs and expressions require manual repair. The length/character restriction is a local supported subset. Native input is:

{
  "parameterSchemaVersion": 1,
  "migrationAcknowledged": true,
  "rawRetentionAcknowledged": true,
  "onePhysicalAttemptAcknowledged": true,
  "videoId": "dQw4w9WgXcQ",
  "part": "snippet"
}

Deliberately select the three initially unchecked acknowledgments after reviewing retention and one-attempt limits. The request is exactly one GET https://www.googleapis.com:443/youtube/v3/videos?part=snippet&id={encodedVideoId} with the selected supplied Bearer token. The project must enable YouTube Data API v3 and have quota and appropriate private-data grants. Original defaults (channel, getAll, wildcard parts), extra options, upload/write/getAll, other resources, Tool/Trigger, token acquisition/refresh, pagination, retries, redirects and returned-URL fetch remain unsupported. Unknown original intent is visibly refused and preserved for repair.

A coherent successful response returns the original bounded items array, with zero or one matching string ID and snippet. An empty array is success; authorization, HTTP, malformed/duplicate JSON, unsafe numeric values, wrong IDs and overflow are failures. result.data and sourceItemsJson retain the items; for example map result.data[0].snippet.title only when an item exists. Useful unknown nested business fields, including names token, password, __proto__ and constructor, remain data without prototype mutation. Returned URLs remain inert. The raw response envelope is discarded after parsing; a digest and selected safe headers remain. Explicit credential value echoes cause value-free failure. Public items enter ordinary run output/history/downstream mappings and may contain private metadata; this adds no encryption, deletion deadline or scanner guarantee for public results.

The original actor must confirm exact input/account/origin/retention with current direct workspace and organization membership, operator WORKFLOW_LIVE_YOUTUBE and current workspace live preference enabled. One physical claim is written in the same checkpoint CAS as the consumed decision annotation, under the current lease. Changed or missing account/task/wait/decision/effect, cancellation or lease expiry refuses dispatch/output and cannot authorize replay. Local limits: 16KiB input/request, 1MiB body, 4096 reads, depth32, 20,000 JSON nodes, 8MiB output and a total 30-second invocation. Current service, native DNS/TLS execution, restart persistence and Review acceptance remain unverified. See account setup.

X (Twitter v2): one username lookup

The useful native unit is exact ordinary numeric typeVersion: 2 user.searchUser, using one explicit username of 1–15 ASCII letters, digits or underscores. It sends one GET https://api.x.com/2/users/by/username/{username} with no query or body. The current documented hostname differs from original api.twitter.com; support for that old alias remains unverified, and redirects are refused. Original default tweet/resource conversion, v1/OAuth1, me/ID lookup, writes, extra fields, expansions, multiuser, paging and retries are visibly unsupported. Import retains original source ledgers and requires explicit repair and account reselection.

In Connections, choose X (Twitter v2) and enter your existing bearer in the named accessToken field, or Advanced JSON {"profile":"xUserV2ManualBearerV1","accessToken":"YOUR_EXISTING_TOKEN"}. Save/update/close/reopen use existing encrypted storage, remain unverified and send zero HTTP/DNS. Test is unsupported and bulk Test skips this account. OAuth2 PKCE acquisition/refresh and original OAuth or broad-scope aliases are not provided. Current X endpoint authorization and credits/pay-per-use require a suitable account; actual grants, entitlement, credits and visibility remain unverified without a real provider test.

In Builder, select user.searchUser, the exact saved workspace account and the username, and explicitly check migrationAcknowledged, outputRetentionAcknowledged and onePhysicalAttemptAcknowledged with schema version2. Current workspace live preference AND the operator WORKFLOW_LIVE_TWITTER gate are required. The original actor needs current organization AND direct workspace execution membership and confirms the exact material in a separate Human Task. Current account bytes, identity, original wait/decision, workflow/version/input, consumed resume segment and nonempty owned lease are checked before the physical claim, HTTP and output persistence. Rotation, deletion, retyping, deauthorization, membership/live changes, later resume/retry/recovery, removed/tampered claims and expired leases refuse further dispatch; uncertain outcomes never automatically replay.

The bounded original data user object is returned as data and one original-style item in sourceItemsJson. String snowflake data.id keeps precision. data.name, data.username and bounded unknown nested business fields are available to ordinary output, run history and downstream mappings, and may contain personal information. The raw response envelope and unknown outer fields are discarded; no plaintext raw artifact or new encryption purpose is introduced. Returned URLs remain inert. Explicit credential VALUES (including encoded forms) are refused on reflection; unrelated business field names are retained. Not-found, rate-limit, credits, auth, duplicate keys, malformed/unsafe values and partial errors fail instead of returning an empty user.

Local bounds are one physical attempt, total30s, response1MiB, depth32 and20k nodes. These are local bounds, not vendor quotas or a deletion/scanning guarantee. Current provider service, Review organization walkthrough, native DNS/TLS and durable DB/Redis restart are untested; local HTTP/Prisma/DOM simulations do not establish them.

PayPal: one payout item read

The native partial operation is ordinary n8n-nodes-base.payPal, numeric typeVersion: 1, payoutItem.get. It sends exactly one GET /v1/payments/payouts-item/{payoutItemId} on the saved explicit live (https://api-m.paypal.com:443) or sandbox (https://api-m.sandbox.paypal.com:443) environment. Enter one explicit string ID; numeric coercion, paths, query controls and returned links are refused.

Save a dedicated PayPal account through Named fields or Advanced JSON:

{
  "profile": "paypalPayoutItemManualBearerV1",
  "environment": "sandbox",
  "accessToken": "YOUR_CURRENT_MANUALLY_SUPPLIED_TOKEN"
}

Both editors use the same central validator. Save, update, close and reopen are local encrypted, write-only and unverified, with zero provider HTTP or DNS. Test is unavailable and bulk Test skips this profile. The original payPalApi client ID/secret credential and its hidden authorization POST are separate and unsupported. No OAuth acquisition or refresh occurs; replace an expired manual token yourself. sanbox, omitted environments and arbitrary fallbacks are refused. PayPal Payouts enablement, token expiry, merchant transaction visibility, current permissions and quota remain unknown without real account verification. The application's organization is distinct from the PayPal merchant identity.

Use parameterSchemaVersion: 2, an explicit payoutItemId, and independently check migrationAcknowledged, outputRetentionAcknowledged and onePhysicalAttemptAcknowledged. These acknowledgments start unset. Imported unsupported resources, versions, extra parameters and original credential references remain visible in the source ledger for manual repair and explicit account reselection. Only the original actor with current direct workspace and organization membership can approve the exact account, environment, route and retention in a separate Human Task. Current saved account material, live environment gate WORKFLOW_LIVE_PAYPAL, workspace preference, original task/wait/decision, resume segment and lease are rechecked at claim, dispatch, settlement and output admission. A removed or changed effect record, expired lease, deletion, rotation, retyping, deauthorization or lost membership cannot authorize a resend or account fallback.

The useful single payout-item object appears in result.data and the one-element result.items array. It retains string identifiers, nullable values and nested unknown business fields, including recipient and financial data, in ordinary execution history and downstream mappings. Review those fields before sending them to another action or AI step. The bounded raw HTTP text and transport envelope are discarded after parsing; no encrypted raw-response archive, new TTL, deletion or scanner guarantee is added. Exact credential values echoed in any key or value are refused; business fields are not removed because of their names. Provider errors and not-found responses are failures, while a payout item's documented business status such as FAILED is preserved as business data.

One consumed decision permits one physical request with a total real 30-second deadline, 1 MiB response, depth 32, 20,000-node and finite-read limits. Duplicate JSON keys, invalid UTF-8/media/status, numeric precision loss, overflow, redirects and ambiguous outcomes fail closed with no automatic retry. Payout create/get/list, payout-item cancel, other resources, triggers, dynamic loaders, pagination, returned-link fetching and token acquisition/refresh remain unsupported. This one selected read is not complete PayPal parity. Tests use simulated HTTP/Prisma/DOM facts; real PayPal, native DNS/TLS, a durable database restart, real AI interpretation and the matching authenticated Review workspace/organization tutorial walkthrough remain untested.

Primary endpoint contract: PayPal payout item details.

Imported SMTP mail

The existing mail action accepts From, To, CC, BCC, Subject, Text and HTML with a saved SMTP account. Ordinary n8n emailSend imports require explicit source version 1, 2 or 2.1. V1 reads CC/BCC at the top level; V2/2.1 read them from options and honor the selected email format. Unsupported attachments, Reply To, attribution, altered TLS/host-name semantics and response-wait operations require repair before preparation or sending. See the import and repair steps. Import never grants send approval.

Firestore: one typed document page

Google Cloud Firestore → List one typed document is the ordinary numeric n8n version 1/1.1 document/getAll subset with explicit returnAll:false, simple:false and limit:1. Supply a Google project ID, explicit (default) database and one ASCII collection ID (letters, digits, underscore or hyphen, at most 128 characters). One input is supported. Nested parents, other databases, defaults, get/batchGet, queries, writes, Tool/Trigger variants and full-node parity are not included. Unsupported original intent, raw versions and credential IDs remain visible for manual repair; reselect the exact typed saved account.

Set parameterSchemaVersion:1 and explicitly select the initially unchecked migrationAcknowledged, rawRetentionAcknowledged and onePhysicalAttemptAcknowledged controls. Imported string "true", numbers, null and expressions never supply consent; unrelated edits and export/reopen retain invalid original values until deliberate repair. The three current stored acknowledgment values themselves must be literal true; business-field expressions do not supply consent. The workspace Firestore live preference, when explicitly boolean, overrides WORKFLOW_LIVE_FIRESTORE; otherwise exact environment false disables and an absent value enables. These controls never replace account or run authority.

The original actor needs current direct workspace and organization membership and independently confirms the exact account/revision, project, collection, route and retention in a separate Human Task. The native host rechecks the current actor, account/secret revision, node/input, live policy, original task/wait/stable decision, consumed resume revision/checkpoint attempt and nonempty owned lease. One checkpoint compare-and-swap records the physical claim and consumed decision annotation together before HTTP. A removed, renamed, uncertain or already used claim cannot authorize another request, including two fresh resumes. The response and fresh OUTPUT require the same current bindings and authenticated consumed receipt.

The single request is GET https://firestore.googleapis.com/v1/projects/{projectId}/databases/(default)/documents/{collection}?pageSize=1, using the manually supplied datastore bearer. Native public-IP admission, reserved TLS/SNI and the fixed public HTTPS443 origin apply. Private and IPv6-only targets, redirects, retries, pagination, token/URL following and extra query/body controls are refused. Local limits are a total thirty-second host/client deadline, an eighteen-second native transport sub-budget, 64 KiB response, depth 24, 4,096 value nodes and 4,096 reader steps.

result.data retains a typed ProtoJSON page with documents, nextPageToken, pageOnly, hasMore and raw. Int64 values stay strings; bytes, references, nanosecond timestamps, maps, arrays, null, false and zero remain distinct. Useful unknown business fields, including __proto__, constructor and prototype, stay inert data. The raw page envelope/token is retained in ordinary history and downstream mappings. This differs from n8n's derived id, item flattening and discarded page envelope; no derived ID or simplification is added. Omitted repeated documents produces an empty page. Duplicate keys, malformed UTF-8/JSON, invalid typed unions/base64, unsafe unknown numeric integers, wrong document ownership/status/media, bounds violations and credential-value reflection refuse success.

Actual provider acceptance, IAM versus Rules, API/database setup, quota, token claims, native deployment, durable database/Redis restart and authenticated Review workspace/organization/tutorial walkthroughs remain unverified.

The three saved acknowledgment properties themselves must be literal true, independently of resolved request or trigger input. Expressions cannot grant consent, including restoration of an older signed WAIT; business fields retain their existing expression support. Invalid original or draft values remain unchanged through export/reopen until deliberate repair.

Additional native Firestore actions

The same selected manual datastore token also supports the following explicitly configured native actions in the (default) database. Each uses the existing three literal migration/retention/one-attempt acknowledgments plus a separate original-actor exact-request Human Task. Current organization AND workspace membership, account revision, input, policy, owned lease, one physical CAS and fresh OUTPUT checks still apply.

No item/column inference, derived document IDs or simplification is added. Typed values preserve int64 strings, bytes, references and timestamps; unknown business metadata remains inert. Writes/deletes are billed and may take effect even if their response is lost; they never retry automatically. Repeating a service-assigned create manually can create another document. Request bodies are limited to8KiB; existing64KiB response, depth24,4,096-node/read,30-second total and18-second transport bounds remain. Returned tokens and URLs are never followed. IAM, Rules, quota and real effects remain unverified.

For IN and NOT_IN, the outer arrayValue.values is a list of comparison operands; an operand may itself be an array value, as in Firestore's array-field query example. Each operand retains the typed-value and selected-project checks. Stored document values, cursors and other filter values still reject directly nested arrays.

The six additional actions require manual migration: pinned n8n output item shapes and item-derived write fields/IDs differ. Import keeps original intent and a specific repair reason instead of silently changing semantics; the existing narrow document.getAll import remains available. Select the account and acknowledgments deliberately after repair.

Primary contracts: Firestore REST documents, structured queries.

Microsoft Dynamics: one Dataverse account read

The native Dynamics profile supports account.get, account.getAll, account.create, account.update and account.delete on the saved public commercial HTTPS {environment}.crm.dynamics.com origin at port443, using /api/data/v9.2/accounts. Other regions, sovereign/on-premises endpoints, other tables, lookups, loaders and Tool/Trigger variants remain unsupported.

Save the dedicated dynamicsAccountManualAccessTokenV1 profile in Named fields or Advanced JSON with accessToken, exact environmentOrigin, matching declared audience and Unix-millisecond expiresAtMs. Save/update/reopen are encrypted, write-only, local and unverified with zero provider HTTP/DNS. Test and OAuth acquisition/refresh are unavailable; historical microsoftDynamicsOAuth2Api credentials are never converted. Declarations do not verify real token grants, tenant consent, licensing or security roles. At least30 seconds of declared lifetime must remain at execution.

Create/update accept the supported text, Int32, four-decimal Money, stage GUID and up to two address rows. Other columns, deep inserts and lookup bindings require another integration. Dataverse plug-ins, flows, business rules and auditing may run for writes. HTTP acceptance does not verify their completion; an uncertain write may already have occurred and never retries automatically.

Set parameterSchemaVersion:1, then independently select literal migrationAcknowledged, rawRetentionAcknowledged and onePhysicalAttemptAcknowledged. Create/update additionally require consequentialEffectAcknowledged; delete requires destructiveAcknowledged. All start unset. Expressions cannot grant consent. The original actor with current direct workspace AND organization membership must approve the exact account, method, body, retention and effects in a separate Human Task. Current account/secret, node/input, live policy, original decision, owned lease/checkpoint/native resume and one physical CAS claim bind dispatch and fresh OUTPUT. Used, expired or uncertain claims never resend.

Raw JSON, numeric lexemes, bounded records and unknown business fields enter normal history and downstream mappings. Links remain inert. Local caps remain30 seconds, 1MiB response, depth32,20,000 nodes,4,096 reads and6MiB output; malformed UTF-8/JSON, duplicate keys, invalid status/media or identity mismatch fail closed. Documented application/json; odata.metadata=minimal is accepted. Real Dataverse rights, provider effects and durable restart behavior remain unverified; local provider/socket/database tests are simulated.

Automatic import accepts only exact ordinary numeric version1 get/getAll/create/delete with explicit supported controls; Return All is refused. Original update requires manual migration because of its upsert semantics. Source ledgers retain parameters, versions and credential references without selecting an account or granting consent; export/reopen preserves invalid drafts until deliberate repair.

Primary contracts: retrieve a row, create a row, update and delete.

GA4 one-row raw report

GA4 → report.get supports one explicit ordinary Google Analytics version2 GA4 report, limited to one row. Choose propertyType: "ga4", a positive decimal property ID with locator mode id, dateRange: "custom", explicit valid YYYY-MM-DD dates, exactly totalUsers, zero or one date dimension, returnAll: false, simple: false, and numeric limit: 1. Dates from 2000 through 9999 and the 20-digit ID bound are local limits. Only the optional boolean keepEmptyRows and returnPropertyQuota flags are accepted; false flags are omitted from the request. Unknown filters, ordering, currencies, metrics or options require repair.

Review the three initially unchecked migration, raw-retention and one-physical-attempt acknowledgments, and select the exact saved GA4 token. Quoted booleans and other invalid original values remain visible and unaccepted. The original actor must hold current direct workspace AND organization membership and approve the exact account, POST body and retention in a separate Human Task. One owned lease/checkpoint segment and physical CAS claim bind dispatch, response persistence and fresh OUTPUT; a used, expired or uncertain attempt never resends, including fresh resumes.

The request is one fixed POST to https://analyticsdata.googleapis.com/v1beta/properties/{id}:runReport, with string wire limit "1". result.data.report and result.sourceReportJson retain the bounded raw report and original JSON in normal history and downstream mappings. Metric/dimension values remain strings, unknown business fields including prototype-shaped keys remain inert data, and unknown numeric lexemes remain in original JSON while the convenience parsed view uses ordinary JavaScript numbers. rowCount, metadata, thresholding, sampling and quota fields never establish complete data. No automatic pagination, retries, redirects or returned URL fetching occurs. Local transport bounds include a public IPv4 TLS/SNI connection, 256KiB, 4096 reads, fatal UTF-8 and 30 seconds. Real provider acceptance, native DNS/TLS, durable database/restart behavior and Review remain untested; local tests simulate provider and database facts.

The three saved acknowledgment properties themselves must be literal true, independently of resolved request or trigger input. Expressions cannot grant consent, including restoration of an older signed WAIT; business fields retain their existing expression support. Invalid original or draft values remain unchanged through export/reopen until deliberate repair.

GA4 bounded report.run

report.run extends the same manual-token account with up to10 metrics and9 dimensions, listed or explicit API names, custom expression/invisible metrics, one AND/OR group of supported dimension and metric filters, ordering, currency, aggregates, empty-row and quota flags. Use returnAll:false and explicit limit1–1000; the requested rows×columns must also fit the10,000-node structural budget. One8KiB POST body is bound to the Human approval; no offset or subsequent page is requested. Data API token/quota consumption and metric/filter compatibility remain unverified.

Custom dates accept YYYY-MM-DD or offset-free local date-time and send the date part. Native today/yesterday/last7days/last30days use Google's property time zone, which can differ from n8n's host-clock dates. simple:true adds string-valued simpleItems inside one retained output envelope, rather than emitting n8n's separate row items. Raw report and original JSON always remain in normal history/downstream values, including unknown inert business fields. Column identities, row counts and string metric values are validated; rowCount never proves completeness. The existing256KiB/depth16/10,000-node/4,096-read/30-second limits, no-retry policy and current organization/workspace authority checks remain.

Automatic import of the broader numeric version2 ordinary GA4 source requires explicit custom dates and simple:false. Relative presets and simple row items retain a manual-migration blocker because those semantics differ. Source credentials, cached locator URLs and metadata never grant account authority or fetch another URL. Reselect the exact saved token, deliberately check all three initially unset literal acknowledgments, and complete the separate original-actor Human Task. The narrow report.get contract remains available. Universal Analytics, Tool/Trigger, OAuth acquisition/refresh and full-node parity remain unsupported. Real Google access and durable restart behavior remain unverified; local execution tests use simulated provider/database facts.

Primary contract: Data API runReport.

Generic S3-compatible file page

The separate generic-s3 action file.getAll.firstPage reads one bounded ListObjectsV2 page from an explicitly supplied custom endpoint. It does not use the AWS S3 bucket account or its regional endpoint rules.

Save genericS3ManualSigV4V1 with profile generic-s3-manual-sigv4-v1, a public DNS HTTPS root origin on port 443, a supplied signing region and an explicit pathStyle boolean. Named fields require an addressing-mode selection; Advanced JSON requires a literal boolean. Save, update and reopen are local, encrypted, write-only and unverified, with zero provider HTTP or DNS. Test and bulk Test are unavailable. Manual-static supplies access and secret keys without token/expiry; manual-temporary also requires a token and exact UTC expiry. Execution needs more than ninety seconds remaining. No ambient credentials, discovery, login, token acquisition or refresh occurs.

Choose an explicit bucket, prefix and limit 1–1000. The migration, first-page and data-retention acknowledgments start unchecked and require literal true, followed by the separate original-actor Human approval. The current actor, workspace and organization access, saved account revision, node, input, live policy, original task/wait/decision and owned execution lease stay bound through one claim, native TLS/SNI/IP admission, the response and fresh output. A used or uncertain claim cannot resend.

One signed GET uses list-type=2, max-keys, delimiter=/ and the exact prefix. It performs no bucket-location request, redirect, retry, continuation or write. Local limits are 1 MiB, 64 response reads and a 30-second total request budget. The bounded XML decoder preserves zero-size and folder-marker objects, keys and unknown business elements; sizes remain decimal strings. The inert XML subset permits at most 32 levels and 10,000 elements, and refuses DTD, external/custom entities, CDATA, custom processing instructions and non-namespace attributes. Optional KeyCount remains provider-declared metadata; actual object and common-prefix counts are bounded independently. partialPage:true, pageCount:1, isTruncated and an optional opaque continuation token describe this first response only. Returned keys and tokens stay inert and are never fetched. Data may persist in run/Human history and reach downstream AI.

This is a corrected ordinary first-page subset. Original generic S3 discovers bucket location, paginates and filters zero-size objects; those behaviors are deliberately excluded. Custom vendor compatibility, entitlement, real DNS/TLS/provider responses, durable database behavior and Review workspace/organization Help walkthroughs remain unverified by local simulated tests. See ListObjectsV2.

Original n8n expression syntax in bucket or folder-prefix fields requires manual repair and stays in the source ledger. Deliberately authored native business expressions use Confluye semantics; they cannot supply the three literal acknowledgments.

Redis explicit string key

redis/string.get (redis_string_get) reads one explicit string key from a selected redisManualTlsV1 external TLS account. Supply numeric parameterSchemaVersion:2, numeric sourceVersion:1, key, literal propertyName and actual booleans migrationAcknowledged:true, rawRetentionAcknowledged:true, onePhysicalAttemptAcknowledged:true. Each consent switch starts unset; imported strings, numbers, null or objects remain invalid and unchecked until explicitly repaired. Saving or reopening another field does not accept them.

The original actor with current organization AND workspace authority approves the exact resolved request in a separate Human Task. One leased CAS claim permits explicit AUTH/SELECT and one GET, with no automatic retry or replay. data.value preserves text; data.missing:true means missing null, while empty string remains empty. data.output uses the literal property name, without dot notation. WRONGTYPE fails. Values enter ordinary history and downstream mappings.

The native owner bounds DNS/TLS/auth/read/close to thirty seconds, one MiB, 128 reads and 64 KiB chunks. Public IPv4 and verified TLS only; no HTTP gateway, queue credential, enumeration, TYPE, write, cluster redirect, reconnect or offline queue substitutes for this operation. Actual Redis/native network and Review remain untested. See the account and limits.

Xero: one existing-tenant contact read

The native ordinary numeric version1 subset is contact.get: one GET https://api.xero.com/api.xro/2.0/Contacts/{contactId} with an explicit UUID and Xero-tenant-id. Save the exact xeroContactManualTokenV1 JSON profile with tokenKind: "oauth2-access-token", manually supplied accessToken, existing tenantId and granted contactsScope (accounting.contacts.read or accounting.contacts). The external Xero tenant is distinct from the current application organization. Named fields and Advanced JSON Save/update/reopen are encrypted, write-only, local, unverified and send zero HTTP/DNS; Test and bulk Test skip. No OAuth refresh or tenant discovery.

All seven explicit boolean migration, output-retention, one-attempt, selected-account, tenant-authorization, approved-usecase and associated-user data-passthrough acknowledgments start unchecked and must be literal true. These user attestations never prove Xero approval or activate Live. The original actor separately decides the exact Human task; current organization AND workspace membership, account/material, node/input/policy, consumed decision, owned lease and one CAS physical claim are revalidated through persistence and fresh OUTPUT. Current provider approval, token/tenant entitlement and associated-user consent remain unverified; resolve those before actual provider activation/testing. Current Xero terms restrict approved use cases, third-party passthrough and model training/adaptation; they do not establish a blanket inference ban.

One exact contact in data and sourceItemsJson preserves bounded original business JSON, including normal prototype-shaped keys and possible personal, bank or tax information, in ordinary output/history/downstream. The raw envelope is discarded; returned URLs stay inert. Local30s/1MiB/fatalUTF8/structural bounds, zero redirects/retries and consumed receipts prevent automatic replay of uncertain attempts. Import/export/reopen retain unsupported raw intent, raw versions and credential IDs; no automatic consent or OAuth-credential alias. Invoice, list, writes, Tool/Trigger and fullnode parity remain unsupported. Real provider, DNS/TLS, Review, durable DB/restart and real human authentication remain untested.

Consent must be saved as seven literal true booleans; acknowledgment expressions are unavailable and cannot acquire consent by resolving to true. A timely settled contact can pass through a later Wait under a new trusted resume revision and current owned lease, with its original signed task/wait/decision/claim/result conserved and current authority checked again; OUTPUT never sends again. The original send approval expires after15minutes; a timely signed settled result may outlive that send window, while retries, recovery or changed checkpoint attempts refuse.

Azure Storage: first container page

Use Azure Storage → List containers (first page), operation container.getAll / tool azureStorage_container_getAll. This is an ordinary numeric n8n typeVersion 1 compatibility subset. Save an existing account in Named fields or Advanced JSON:

{
  "profile": "azureStorageManualBearerV1",
  "account": "youraccount",
  "accessToken": "YOUR_EXISTING_STORAGE_BEARER",
  "audience": "https://storage.azure.com/",
  "expiresAt": "2099-01-01T00:00:00.000Z"
}

Use the token's actual expiry; the example date is a placeholder. Account is 3–24 lowercase letters/digits, without a URL. Save/update/reopen encrypt locally, are write-only and remain unverified with zero provider HTTP. There is no Test, OAuth acquisition/refresh, SharedKey signer, SAS, discovery or Azure resource creation. Token signature/issuer/audience/expiry claims and effective rights remain unverified. List Containers permissions require Microsoft.Storage/storageAccounts/blobServices/containers/read scoped to the account or above; a blob-level entitlement alone does not prove account-list access. Tokens expiring within 30 seconds are refused at execution.

Original imports must declare exactly ordinary numeric version 1, resource:container, operation:getAll, returnAll:false, limit:1, and matching original authentication:oAuth2/azureStorageOAuth2Api or authentication:sharedKey/azureStorageSharedKeyApi. Original declarations and unsupported intent remain in the source ledger. Reselect the separate manual bearer account; no original OAuth/SharedKey authentication parity is claimed. Other defaults/options, metadata/system/deleted filters and Tool/Trigger require visible manual repair; additional admitted operations are described below.

Input requires parameterSchemaVersion:2 and four independent literal acknowledgments: migrationAcknowledged:true, rawRetentionAcknowledged:true, onePhysicalAttemptAcknowledged:true and firstPageOnlyAcknowledged:true; none is automatically checked. The original actor needs current direct organization AND workspace membership, current selected account and input revisions, WORKFLOW_LIVE_AZURE_STORAGE=true and the independent current workspace live preference. A separate signed Human Task binds the exact safe request preview without embedding the bearer. Nonempty owned lease, original task/wait/decision, checkpoint attempt and consumed resume revision remain required. One CAS claim permits one physical attempt; uncertain or used claims never resend, including fresh resumes or node retry settings.

The request is exactly GET https://{account}.blob.core.windows.net/?comp=list&maxresults=1, with Bearer authorization, a fresh UTC x-ms-date, and source x-ms-version:2021-12-02. Public DNS, admitted IP, certificate/SNI and remote peer are revalidated. For this first-container-page action, custom/private/sovereign endpoints, alternate ports, redirects, retries and continuation requests are unsupported; IPv6-only hosts remain unsupported.

A successful 200 XML response yields data.containers, data.maxResults:1 and opaque data.nextMarker. paging.completeness is partial-first-page when a marker exists, otherwise complete-first-page; this never implies full pagination. Empty pages retain their marker, fixing the original helper shortcut's loss. Unknown container fields/attributes and textual metadata (including 0007) persist in normal history and downstream mappings. Returned URLs remain inert. Root-level unknown fields are omitted as in the original helper. XML names follow source helper normalization; strict named boolean/safe integer conversion applies outside metadata. Prototype-sensitive names and normalization collisions refuse success. This differs deliberately from the original metadata coercion.

Limits: one total 30-second dispatch/read/decode deadline, 256 KiB body, depth 24, 4,096 elements/reads, one container, and an 8,192-character marker. Strict UTF-8/XML, exact response status/media/length, secret-reflection checks and bounded streaming precede success. DTD, custom/external entities, namespaces, comments, CDATA and processing instructions are rejected. The five predefined XML entities and valid numeric character references are decoded without external resolution. Raw transport text is discarded; no full XML/parser parity is claimed. Existing business data may be private and remains in ordinary history; this adds no retention/encryption guarantee for that history.

Local tests simulate provider responses, Prisma facts and native sockets. Real Azure token claims/entitlement/RBAC propagation, service responses, network ACL/private-network constraints, native deployment, actual durable database/Redis restart and organization-scoped Review/tutorial behavior remain untested.

Azure Storage and Cosmos DB: additional bounded operations

Azure Storage also exposes container.get/create/delete and blob.getAll/get/create/delete, using the same exact azureStorageManualBearerV1 account and fixed public origin. Select the operation, review its disclosure and supply schema 2 plus literal migration, retention and one-attempt consent. Container creation needs createContainerAcknowledged; container deletion needs deleteContainerAcknowledged and removes every blob. Blob upload needs replaceBlobAcknowledged, explicit UTF-8 content (at most 4 KiB) and contentType; it can replace content and metadata. Blob deletion needs deleteBlobAcknowledged; snapshots, leases and conditional requests are unsupported. Every write requires a separate exact-request Human Task and one physical claim; a lost response never authorizes retry.

blob.getAll requires firstPageOnlyAcknowledged and an explicit limit of 1–100. Optional include accepts only the fields exposed by the form. It returns data.blobs, data.maxResults and an opaque data.nextMarker; empty partial pages retain their marker and it is never followed. blob.get requires textOnlyAcknowledged and returns bounded UTF-8 data.content with name, properties and metadata. Original n8n blob downloads produce binary items and remain refused; configure this different text-only operation manually. Original blob uploads using binary or URL sources also remain refused. Container read/create/delete, bounded blob listing and deletion can migrate only their admitted literal source options. Unsupported fields remain visible for repair. XML keeps the 256 KiB/depth-24/4,096-element bounds; text download is at most 256 KiB. All calls retain the total 30-second budget.

Azure Cosmos DB also exposes container.getAll/create/delete and item.get/getAll/query/create/update/delete with the exact azureCosmosDbMasterKeyV1 account, saved database and fixed public origin. Every operation requires schema 2 and literal migration, retention and one-attempt consent; all five writes additionally require writeEffectAcknowledged. The original actor needs current organization AND workspace membership and both live gates, then confirms the exact selected account, request and body. Every action uses one HMAC-authenticated request and one physical claim; uncertainty remains unverified and never resends.

Container/item listing accepts one page with maxItemCount 1–100; continuation is opaque and never followed. Query sends one POST with explicit SQL and named parameters, not a write. Choose crossPartition (potentially higher RU charges) or singlePartition with an explicit string partition key. It accepts at most 1,000 returned values; a page may be empty or partial. Point item reads and writes need an explicit string partition key; no discovery probe or inferred key is used. Return values are data.containers, data.items, data.item, data.container or {deleted:true}, with bounded envelopes and numeric-lexeme metadata. New item/query responses preserve otherwise unsafe numeric spellings as text. The original container.get contract retains its stricter numeric refusal. Normal history and downstream mappings retain business data.

Native item creation never upserts; an existing id conflicts. Native item update replaces the whole item using the exact authored JSON and matching id; omitted properties disappear. Both differ from original n8n upsert/injected-body behavior, so those imports remain refused and require manual configuration. Container creation accepts one Hash partition path and optional indexing policy/manual throughput: 400–1,000,000 RU/s in steps of 100 can incur ongoing billing. Deleting a container removes all contents. No patch/upsert/autoscale, retries, redirects or automatic continuation is implemented. Limits remain total 30 seconds, 1 MiB response, depth 32 and 20,000 nodes; authored item/policy JSON is at most 8 KiB.

These are bounded native operations and explicit migrations, not full n8n item/authentication parity. Local HOST tests use simulated provider responses and database facts. Real provider credentials, grants, charges, native DNS/TLS, durable restart and authenticated organization Review/tutorial behavior remain untested.

Firebase Realtime Database one-path read

Partial support is exact ordinary n8n-nodes-base.googleFirebaseRealtimeDatabase numeric typeVersion:1, explicit operation:"get", a supplied database instance and one literal nonroot path. The native action is get. Select the exact saved firebaseRealtimeManualBearerV1 workspace account; databaseInstance must match its origin's instance label. Set schema1 and deliberately check migrationAcknowledged, rawRetentionAcknowledged, and onePhysicalAttemptAcknowledged; all three start unchecked on import. pathSegments accepts 1–32 nonempty ASCII alphanumeric/underscore/hyphen keys, each at most128 bytes. This is a local subset, not every legal Firebase key.

One GET https://<instance>.<supported-suffix>/<joined-path>.json carries no body or query. One selected path can expose its entire subtree and consume quota, bandwidth or charges. The bounded output keeps data, valueType and exact rawJson: false, zero, empty string, null, arrays and objects remain distinct; no exists inference or n8n scalar wrapping/null-exception parity is promised. Parsed decimals use JavaScript number precision; rawJson preserves lexical values and unsafe integers/nonfinite values refuse. Unknown fields and URLs remain inert data in ordinary workflow history and downstream mappings. Credential value echoes fail without retaining values. No new history encryption, TTL or scanner guarantee is added.

Both live controls (WORKFLOW_LIVE_FIREBASE and workspace preference), current direct workspace AND organization membership, exact current account/input/workflow/version, original-actor Human decision, matching consumed task/wait/stable decision, nonempty owned lease and one checkpoint CAS claim are required. Current authority is checked again before HTTP and result persistence; missing/tampered/ambiguous outcomes never resend, including two fresh resumes. Local bounds: total30s, body1MiB, depth32,20k JSON nodes,4096 stream reads and one attempt; no redirect/retry. Root/discovery, filters/query/shallow/export/SSE, pagination, Tool/Trigger, unsupported keys and OAuth refresh remain refused. Real provider/Review/native sockets and durable DB/Redis restart are untested.

Firebase acknowledgment controls accept real booleans only. Imported strings, numbers, null and objects remain visibly invalid and are retained unchanged on unrelated edits; choosing each unchecked switch is a separate user action. No expression or default turns an invalid value into consent.

For every native operation, restoring a signed saved wait still requires the current source acknowledgment fields to be literal true; neither a restored signature nor a resolved expression supplies consent. This check does not change business-field expression rendering.

Explicit ordinary numeric version 1 Firebase update, push, create and delete now migrate to separate bounded native writes: PATCH, POST, PUT and DELETE respectively, using the same firebaseRealtimeManualBearerV1 account and one nonroot literal path. The omitted original default is still refused. Body operations retain the original comma-separated attributes; bind inputData explicitly to the upstream record because n8n implicitly read the current item. Only listed own top-level properties present in that record are sent, as canonical JSON (at most 4 KiB). Requests with no matching attributes, server values or multi-path keys are refused.

All writes require the three common literal acknowledgments plus externalWriteAcknowledged; PUT also needs fullReplacementAcknowledged because omitted subtree values are removed, and DELETE needs subtreeDeletionAcknowledged because it removes the entire subtree. Import leaves every consent false and does not select or acquire credentials. Review the exact account, verb, path and body in the original-actor Human Task under current organization AND workspace authority and both live gates. One durable physical claim permits one attempt; a timeout, disconnect or malformed response leaves the outcome unverified with no automatic retry or replay. Successful writes retain bounded data, rawJson and request-body byte count in ordinary history. POST returns the generated child name; DELETE returns null. These are explicit adaptations; original scalar/null/item wrapping and implicit per-item execution parity are not claimed. Real provider writes, charges and Review walkthroughs remain untested.

QuickBooks: one explicit bill

Select exact ordinary numeric-v1 bill.get, one string billId and the saved quickbooksBillManualBearerV1 account. Its explicit production/sandbox realm determines the fixed Intuit origin. IDs use a local 1–40 ASCII decimal-string subset, preserving leading zeros; it is not the complete provider grammar. The initially unchecked acknowledgeFinancialDataPersistence must be literal true, including on restored WAIT; expressions/truthy strings cannot approve retention. The original actor separately confirms the Human task under current organization AND workspace/account/decision/lease/checkpoint authority and one physical CAS claim.

One GET uses minorversion=75, a floor/default rather than a frozen complete schema, with one thirty-second total deadline, bounded fatal-UTF8 JSON and no retry/redirect. The exact string-ID open Bill and exact rawResponseJson retain unknown fields and financial number text in ordinary history/downstream data; parsed decimals are not exact financial arithmetic. Fault, missing/mismatched Bill, duplicate keys, unsafe integers and credential echoes fail. Consumed or uncertain claims never resend; fresh OUTPUT rechecks the settled receipt. See account fields, refusals and limitations. Other actions/options, list/write, Tool/Trigger, OAuth refresh and complete Bill/node parity are unsupported. Real provider/native network, durable restart and Review remain untested.

CrateDB and TimescaleDB bounded read

Use CrateDB / TimescaleDB bounded read → Read one reviewed row, provider native-pg-read, operation query.execute, tool native_pg_read_query_execute. Select the explicit nativePgReadManualTlsV1 account with a matching provider profile and reviewed schema, table and projected columns. This uses PostgreSQL wire transport over verified system-trust TLS; it is a separate action from legacy postgres-query.

The exact input is parameterSchemaVersion:1, sourceVersion:1, profile:"crateDb"|"timescaleDb", literal query, sourceMode:"multiple"|"independently", and migrationAcknowledged, rawRetentionAcknowledged, onePhysicalAttemptAcknowledged. Each acknowledgment starts unchecked and its saved raw property must be literal true before expression resolution. Exactly one actual original input item and a separate exact original-actor Human approval are prerequisites. Current actor, account revision, node/input, workspace and organization authority, live policy and owned attempt remain bound through execution and output.

Only SELECT column [, column]* FROM schema.table [ORDER BY one projected column [ASC|DESC]] LIMIT 1 is admitted. SQL is retained and sent exactly, with no rewrite or implicit setup query. No parameters, WHERE, star, alias, expressions, functions, comments, joins, subqueries, multiple statements, writes or transaction safety assumption is supported. Limits are 4096 SQL UTF-8 bytes, 256 tokens, 16 columns, one row, 1 MiB wire data, 256 KiB decoded output, 15 seconds total, 5 seconds to connect and a separate 2-second positive cleanup budget. There is one owned physical connection and one query, without retry, address fallback, discovery, pooling or continuation. An uncertain or consumed attempt cannot resend under the same decision.

The explicit native-pg-read-v1 envelope contains items:[{"json":{"id":7,"label":"alpha"},"pairedItem":{"item":0}}] for one returned row and items:[] for zero rows. No empty firstRow is fabricated. Projected business field names remain ordinary data, including names such as password or constructor; account secrets never belong in the result. Only explicitly supported scalar text, integer, boolean, UUID and numeric wire types are admitted. Int8 and numeric values remain lossless strings; unsupported dates, floats, JSON, arrays, binary and OIDs refuse the whole result. attempts:1, outcomeVerified:false and sourceSynthesis:"none" describe the local contract, not verified database semantics. Returned business data persists in ordinary run history and can reach downstream AI. External n8n helper/driver conversion parity, actual server/schema/role behavior, native deployment, durable restart and Review walkthrough remain unverified.

See manual account setup and original import repair.

Lectura acotada en español

Selecciona el perfil y la cuenta manual exactos, la consulta SELECT literal con esquema, tabla y columnas revisados, y un único elemento de entrada. Las tres confirmaciones de migración, retención y un intento deben contener true literal; se requiere otra aprobación Human del actor original con acceso actual al workspace y a la organización. Solo hay una conexión y una consulta sin reintentos ni SQL implícito. Una fila produce un elemento json con pairedItem.item:0; cero filas produce items:[], sin inventar firstRow. Los nombres de campos de negocio se conservan y los datos pueden llegar al historial y a IA posterior. No se afirma paridad completa ni compatibilidad real de proveedor, esquema, permisos, TLS nativo, reinicio duradero o Review.

General node/input revision JSON is bounded to 512 KiB, 20000 JSON nodes and depth 32. The signed approval payload is limited to 32 KiB, and the exact approval preview to 24 KiB. An original-source ledger is limited to 128 KiB (131072 UTF-8 bytes) at execution admission. These limits refuse before an effect; no content is truncated. If positive cleanup cannot be established, no successful output or resend is admitted. The exact execution lease and durable attempt claim are retained, and the unclosed socket remains strongly owned; the host does not assert clean resource disposal. En español: el JSON general de revisiones de nodo y entrada se limita a 512 KiB, 20000 nodos JSON y profundidad 32. La carga firmada de aprobación se limita a 32 KiB y su vista previa exacta a 24 KiB. El registro de origen se limita a 128 KiB (131072 bytes UTF-8) al admitir la ejecución. Un exceso se rechaza antes de producir efectos, sin truncar. Un cierre no confirmado impide éxito y reenvío; se conservan el lease exacto de ejecución y la reserva duradera, y el socket sigue bajo propiedad fuerte.

Protocol frame sublimits also refuse the whole result: 8 KiB for authentication/metadata/notice/error frames, 1314 bytes for a row description and 256 KiB for a DataRow. Quoted ASCII identifiers are required for reserved or special keywords. En español: también se rechaza el resultado completo si excede 8 KiB en tramas de autenticación/metadatos/avisos/errores, 1314 bytes en la descripción de fila o 256 KiB en DataRow; las palabras reservadas o especiales requieren identificadores ASCII entre comillas.

LDAP bounded base search

Choose LDAP bounded base search → Search one exact LDAP base DN, provider native-ldap-read, operation search.base, tool native_ldap_read_search_base. The saved nativeLdapReadManualLdapsV1 selection must match exactly. Input is:

{
  "parameterSchemaVersion": 1,
  "sourceVersion": 1,
  "selection": {
    "baseDN": "cn=target",
    "scope": "base",
    "returnAll": false,
    "limit": 1,
    "paged": false,
    "attributes": ["cn", "mail"],
    "filter": {
      "kind": "equality",
      "attribute": "cn",
      "value": "target"
    }
  },
  "migrationAcknowledged": false,
  "rawRetentionAcknowledged": false,
  "onePhysicalAttemptAcknowledged": false
}

Review each acknowledgment and explicitly change its saved value to literal true. Expressions cannot grant consent. One actual original input and separate exact Human approval by the original actor are required, with current organization AND workspace authority, saved credential revision, workflow/input/live-policy revisions, owned lease and one durable physical claim. A used or uncertain claim never replays.

One LDAPS connection performs simple bind, one base query and optional unbind. The native-ldap-read-v1 envelope contains operation, result, attempts:1, outcomeVerified:false and a physical receipt reference. result.entry is either null or {dn,attributes:[{name,values}]} from admitted wire facts; no SDK-synthetic missing attributes are published. Leading U+FEFF text is retained. Missing requested attributes are listed. inventoryComplete:false always; empty entry does not prove absence or a complete inventory. Wire code4 remains partial/error, including any retained entry. Other failures and uncertain physical cleanup refuse public output. DNs and directory values persist in ordinary history and may reach downstream AI.

Limits are16 projected attributes,32 values each,8192 UTF-8 bytes per value, one entry,64KiB per BER frame,256KiB ingress/output, eight inbound frames,15 seconds total, five seconds connect and two seconds positive cleanup. Reserved names such as dn, constructor, prototype, __proto__, objectGUID and objectSid, case collisions, binary/invalid UTF-8 and unsupported BER forms refuse the entire result. There is no paging, write, fallback or retry. Required authority checks and synchronous write liveness do not make database/network actions atomic. Actual directory compatibility, production dependency binding/deployment, durable restart and Review remain untested.

En español: search.base exige una entrada real, la selección revisada exacta, tres confirmaciones literales y otra aprobación Human del actor original con acceso vigente a organización Y espacio. El resultado conserva DN, matrices de texto y estados reales del protocolo; código4 sigue siendo parcial/error e inventoryComplete siempre es false. No se fabrican atributos ausentes. Un intento usado o incierto no se reenvía. Los datos se guardan en historial y pueden llegar a IA posterior. Directorio real y Review siguen sin probarse.

MySQL reviewed row

The native-mysql-read operation select, tool native_mysql_read_select, reads zero or one row from one reviewed table through the explicit nativeMySqlReadManualTlsV1 verified-system-TLS account. The supported original is an ordinary numeric n8n MySQL 2.5 select node with one actual input item, literal table locator in name mode, ordered explicit columns, limit:1 and largeNumbersOutput:"text". parameterSchemaVersion:1, sourceVersion:2.5, the intact originalNodeJson, matching table/columns and all three literal acknowledgment properties are required. Default insert, Tool/Trigger, filters, sorting, wildcard columns, raw SQL, expressions, paging, writes and unknown options are unsupported.

The native-mysql-read-v1 envelope contains operation:"select", sourceVersion:2.5, items:[{"json":{"id":"9007199254740993"},"pairedItem":{"item":0}}] or items:[], attempts:1, outcomeVerified:false and sourceSynthesis:"none". BIGINT and DECIMAL stay exact strings; leading Unicode text, null and reserved own business keys remain data. Unsupported or oversized types reject the whole result. Returned rows persist in authorized history and can reach downstream AI.

Limits are 16 columns, one row, 16 KiB per field, 64 KiB per packet/output, 1 MiB wire, 256 KiB plaintext, 15 seconds total, 5 seconds to connect/TLS and 2 seconds for positive cleanup. A MAX_EXECUTION_TIME hint accompanies the finite SELECT; LIMIT does not bound all server work and aborting a client does not prove server cancellation. Current original-actor organization AND workspace/account/input/node/policy, exact Human/lease lineage, durable single claim and fresh signed OUTPUT are required. No used or uncertain attempt automatically repeats. Owned local MySQL 8.4.13 single-factor TLS coverage does not prove a user's privileges, server configuration, public DNS, durable restart or Review walkthrough.

MongoDB bounded document read

Select one actual input, literal ObjectId _id equality, an inclusion projection containing _id plus 1–8 reviewed scalar fields, and limit 1. The exact original query text is { "_id": "0123456789abcdef01234567" }; the selected string is explicitly migrated to a BSON ObjectId. For example, projection text { "_id": 1, "amount": 1 } requires amount in the saved account's reviewed fields. See the complete original node syntax and the manual account.

There is no discovery, write, continuation, retry or fallback. Zero documents emit items: []. A returned document can produce the following scalar Extended JSON v2 item; the Int64 remains a string:

{
  "json": { "_id": { "$oid": "0123456789abcdef01234567" }, "amount": { "$numberLong": "9223372036854775807" } },
  "pairedItem": { "item": 0 }
}

The result wrapper identifies contractVersion:"native-mongodb-document-find-v1", operation:"document.find" and encoding:"canonical-extended-json-v2-scalars". Canonical EJSON scalar output preserves ObjectId, Int64 strings, finite Decimal128, dates and leading BOM; unsupported BSON refuses the whole output. Raw data persist in authorized history and may reach downstream AI. One find may use up to 4 stock-SDK external connections. Bounds are 64 KiB per frame/source, 1 MiB ciphertext, 512 KiB plaintext, 256 KiB output, 15 seconds of work plus 2 seconds for positive cleanup; signed preview is 24 KiB and payload 32 KiB. These are owned-buffer and process lifecycle controls, not a hard OS memory cap. Real provider credentials, deployment packaging and authenticated Review remain unverified.

En español: una entrada y una búsqueda con _id ObjectId, proyección explícita de _id y 1–8 campos revisados, y límite 1. Cero documentos producen cero elementos. El ejemplo EJSON conserva el entero largo como texto sin perder precisión. Los datos se guardan en el historial autorizado. No hay repetición ni descubrimiento. Los límites no constituyen aislamiento duro de memoria del sistema; las pruebas del proveedor, despliegue y Review siguen pendientes.

Jina Reader one page

Choose Jina Reader → Read one public HTTPS page with Jina, provider jina-reader, operation reader.read, tool native_jina_reader_reader_read. Native input is:

{
  "parameterSchemaVersion": 1,
  "sourceVersion": 1,
  "selection": {
    "url": "https://www.wikipedia.org/",
    "simplify": false,
    "options": {}
  },
  "migrationAcknowledged": false,
  "rawRetentionAcknowledged": false,
  "onePhysicalAttemptAcknowledged": false
}

Review the migration, raw-retention and one-physical-attempt disclosures and explicitly save each acknowledgment as literal true. Expressions cannot grant consent. One actual original input, current active original actor, organization AND workspace access, saved-key ciphertext revision, workflow/source/live revisions and separate exact Human approval precede a durable lease/revision CAS claim. The transport pins public service DNS and uses one system-trust TLS connection to fixed r.jina.ai; it never connects directly to the target. Provider redirects, retries, anonymous fallback, query/fragment/IP/special-use targets and escaped/non-ASCII paths refuse.

The jina-reader-v1 output retains requested URL separately from returned data.url, actual HTTP/code/status, the complete bounded open envelope and rawResponseJson. Unknown business fields, false, zero, null, empty content and leading U+FEFF remain intact. Content must be a string; absent data/content or provider errors refuse public output instead of fabricating success. Limits are 256 KiB decoded JSON, depth20/20,000 nodes, 512 KiB output, 15 seconds total/five seconds connect/two seconds positive cleanup. Unconfirmed close retains owned cleanup and is terminal with no replay.

Jina performs the remote target fetch; its target DNS, redirect and access-policy behavior are not local transport facts. One request can disclose the target URL and consume credits. Raw untrusted page data persists in authorized workflow history and may reach downstream AI. Provider/key compatibility, credits, deployment and authenticated Review remain untested. Fresh authority checks and guarded writes do not make database and network actions atomic.

En español: Jina obtiene el destino remoto; su DNS, redirecciones y políticas no son hechos del transporte local. Una solicitud puede revelar la URL y consumir créditos. Los datos sin confianza persisten en historial y pueden llegar a IA posterior. Compatibilidad real, créditos, despliegue y Review autenticado siguen sin probarse. Organización Y espacio, actor original y cuenta guardada vigente siguen siendo obligatorios; un intento usado o incierto no se repite.

AWS Rekognition DetectLabels (preview)

The separate aws-rekognition provider offers rekognition.image.detectLabels for one explicitly versioned S3 image. Execution requires both the deployment setting WORKFLOW_LIVE_NATIVE_REKOGNITION=true and the workspace live preference for aws-rekognition. Human approval belongs to the original actor and binds the saved account, exact source, input and workflow version. Changes require a new approval. An uncertain paid outcome must not be replayed.

Save the exact family awsRekognitionManualSigV4V1 with profile aws-rekognition-manual-sigv4-v1, partition aws, region us-east-1, a declared 12-digit account ID and existing manual keys. manual-static omits sessionToken and expiresAt; manual-temporary requires both a supplied token and UTC expiry. Expired credentials can be saved locally but must be refused at execution within the ninety-second dispatch safety margin. Named fields and Advanced JSON validate the same closed profile. Save/update/reopen remain local, encrypted, write-only and unverified, with no Test, bulk Test, provider request, STS, discovery or refresh. Reopen never fills stored keys; Update account needs a complete replacement and invalidates earlier approvals. A declared account is not verified identity.

The finite source projection accepts only ordinary n8n-nodes-base.awsRekognition numeric version 1, authentication: "iam", resource: "image", operation: "analyze", type: "detectLabels" and binaryData: false. Supply an ordinary S3 bucket, the exact literal object key, an explicit existing non-null S3 version, integer maxLabels from 1 to 20 and finite minConfidence greater than 0 up to 100. Keys retain Unicode and whitespace. Other detectors, binary upload, expressions, omitted defaults, extra fields, automatic retries and continue-on-fail remain unsupported. Imported AWS ID/name references do not select a saved account or grant execution authority.

The request deliberately omits the original Filters: {} and sends explicit Features: ["GENERAL_LABELS"]; acknowledge this migration rather than assuming byte equivalence or full original parity. The S3 bucket must be in the same fixed us-east-1 region as Rekognition. No S3 object fetch, region probe, ownership, principal, object-format, KMS or cross-account verification occurs. A version-bound pointer does not attest the image bytes.

All six acknowledgments start unchecked and require literal true: migrationAcknowledged, billableAttemptAcknowledged, providerRetentionAcknowledged, rawRetentionAcknowledged, versionPointerAcknowledged and sameRegionAcknowledged. These are separate from later Human approval. One physical DetectLabels attempt can incur charges; AWS can continue after local cancellation. A lost result is terminal unknown and must not automatically replay.

AWS may retain or use submitted images unless opted out. The acknowledgment neither verifies nor changes the AWS retention policy or opt-out; review the provider account settings independently. Locally, the exact S3 pointer, complete labels, model metadata and bounded open raw JSON enter ordinary run/Human history and may reach downstream AI. No extra encryption or TTL is promised. Provider labels are observations, not verified truth. The operation bounds the response to 1 MiB and retained output to 4 MiB; JSON Schema exposes structure while Zod additionally enforces literal text, UTF-8 and bucket refinements. Real AWS requests and authenticated organization/workspace Review walkthroughs remain untested. Local simulated responses do not verify AWS permissions, image access or provider retention settings.

Español: DetectLabels en vista previa

El proveedor separado aws-rekognition ofrece rekognition.image.detectLabels para una imagen S3 con versión explícita. La ejecución exige tanto WORKFLOW_LIVE_NATIVE_REKOGNITION=true en el despliegue como la preferencia de ejecución real de aws-rekognition en el workspace. La aprobación humana corresponde al actor original y vincula la cuenta guardada, el origen exacto, la entrada y la versión del flujo. Los cambios exigen una nueva aprobación. Un resultado facturable incierto no debe repetirse.

Guarda exactamente awsRekognitionManualSigV4V1, con perfil aws-rekognition-manual-sigv4-v1, partición aws, región us-east-1, cuenta declarada de 12 dígitos y claves manuales existentes. manual-static omite sessionToken y expiresAt; manual-temporary exige ambos, con caducidad UTC. Guardar credenciales caducadas no las verifica; la ejecución debe rechazarlas dentro del margen de seguridad de noventa segundos. Los campos con nombre y JSON avanzado validan el mismo perfil cerrado. Guardar, actualizar y reabrir son operaciones locales, cifradas, de solo escritura y sin verificar: no realizan pruebas individuales ni masivas, solicitudes al proveedor, STS, descubrimiento ni renovación. Al reabrir, las claves quedan vacías; Actualizar cuenta exige un reemplazo completo e invalida aprobaciones anteriores. La cuenta declarada no acredita identidad.

La proyección acotada solo acepta el nodo ordinario original n8n-nodes-base.awsRekognition con versión numérica 1, authentication: "iam", resource: "image", operation: "analyze", type: "detectLabels" y binaryData: false. Indica un bucket S3 ordinario, clave literal exacta, versión existente explícita distinta de null, maxLabels entero entre 1 y 20 y minConfidence finito mayor que 0 y hasta 100. Se conservan Unicode y espacios de la clave. Otros detectores, carga binaria, expresiones, valores predeterminados omitidos, campos adicionales, reintentos automáticos y continue-on-fail no están admitidos. Las referencias importadas a ID o nombre de AWS no seleccionan cuenta ni conceden autoridad.

Se omite el Filters: {} original y se envía Features: ["GENERAL_LABELS"] explícitamente. Confirma esta migración por separado; no se promete equivalencia de bytes ni paridad completa. El bucket S3 y Rekognition deben estar en la misma región fija us-east-1. No se obtiene el objeto ni se verifican región, propiedad, principal, formato, KMS o reglas entre cuentas. El puntero de versión no acredita los bytes de la imagen.

Las seis confirmaciones empiezan desmarcadas y exigen true literal: migrationAcknowledged, billableAttemptAcknowledged, providerRetentionAcknowledged, rawRetentionAcknowledged, versionPointerAcknowledged y sameRegionAcknowledged. Son independientes de la aprobación humana posterior. Un intento físico de DetectLabels puede generar cargos y AWS puede continuar después de cancelar localmente. Un resultado perdido queda como desconocido terminal y no debe repetirse automáticamente.

AWS puede conservar o usar imágenes enviadas si no se ha optado por excluirlas. La confirmación no verifica ni cambia la política de retención ni la exclusión; revisa la configuración de la cuenta del proveedor por separado. Localmente, el puntero S3 exacto, las etiquetas completas, los metadatos del modelo y el JSON bruto abierto y acotado entran en el historial ordinario de ejecuciones y tareas humanas y pueden llegar a IA posterior. No se promete cifrado adicional ni TTL. Las etiquetas son observaciones del proveedor, no verdades verificadas. La operación limita la respuesta a 1 MiB y la salida retenida a 4 MiB; JSON Schema expone la estructura y Zod añade las restricciones de texto literal, UTF-8 y bucket. Siguen sin probarse las solicitudes reales a AWS y los recorridos autenticados de organización y workspace en Review. Las respuestas locales simuladas no verifican permisos de AWS, acceso a la imagen ni la política de retención del proveedor.

AWS Textract AnalyzeExpense (preview)

AnalyzeExpense requires schema version 1, literal authentication:"iam", operation:"analyzeExpense", binaryPropertyName and simple:false. All five acknowledgment fields must be literal true: migrationAcknowledged, billableAttemptAcknowledged, providerRetentionAcknowledged, rawRetentionAcknowledged and binaryDisclosureAcknowledged. Expressions do not grant consent.

Supply exactly one actual original input item with the selected ready workspace PNG or baseline JPEG. The image limit is 262144 bytes and 4096 pixels per side. The maximum canonical image request is 349553 bytes including base64 and JSON syntax; the request is not truncated. S3 references, inline text, multiple input items and other operations are unsupported. Changing the image, source or account requires new exact Human approval.

Output retains the complete bounded provider object and raw JSON with exact numeric spellings, rather than the original simplified summary mapping. Missing or empty expense arrays remain missing or empty. Unsafe numeric conversion, oversized data and malformed output fail closed. Expense values are unverified provider observations; the feature does not establish financial correctness.

Execution requires both the deployment setting WORKFLOW_LIVE_NATIVE_TEXTRACT=true and the workspace live preference aws-textract. Real AWS requests and authenticated organization/workspace Review walkthroughs remain untested.

Español: AnalyzeExpense en vista previa

AnalyzeExpense exige esquema de versión 1, authentication:"iam", operation:"analyzeExpense", binaryPropertyName literal y simple:false. Las cinco confirmaciones deben ser true literal: migrationAcknowledged, billableAttemptAcknowledged, providerRetentionAcknowledged, rawRetentionAcknowledged y binaryDisclosureAcknowledged. Las expresiones no conceden consentimiento.

Aporta exactamente un elemento original real con el PNG o JPEG baseline listo del workspace elegido. El límite es 262144 bytes y 4096 píxeles por lado. La solicitud canónica máxima ocupa 349553 bytes con base64 y sintaxis JSON; no se trunca. No admite referencias S3, texto en línea, múltiples elementos ni otras operaciones. Cambiar imagen, origen o cuenta exige nueva aprobación Human exacta.

La salida conserva el objeto completo limitado del proveedor y el JSON sin procesar con la escritura exacta de los números, en lugar del resumen simplificado original. Las listas ausentes o vacías siguen así. La conversión numérica insegura, datos excesivos o salida malformada se rechazan. Los gastos son observaciones del proveedor sin verificar; no se garantiza corrección financiera.

Se exige tanto WORKFLOW_LIVE_NATIVE_TEXTRACT=true en el despliegue como la preferencia real aws-textract del workspace. Siguen sin probarse las solicitudes reales AWS y los recorridos autenticados de organización y workspace en Review.

Google language full-response operations

google-translate / translate.n8nV1 takes text, targetLanguage and three literal boolean acknowledgments for text disclosure, possible billing and full-response retention. It omits format and returns the whole first translation object. google-natural-language / analyzeSentiment.n8nInlineV1 takes text and the same three acknowledgments; its inline plain-text/UTF16/autodetect response retains sentences and numeric values. Each choice starts unchecked; independent gateway confirmation remains required. Full text and unknown business fields enter normal history/downstream data within explicit bounds, or the response refuses as a whole. Existing translate and analyzeSentiment contracts remain available. See accounts, consent and limits and strict source import profiles. Live provider and Review flows are unverified.

Google directory raw getters

Choose Google Chat → Get complete bounded space (n8n v1) (spaces.getRaw) or Google Contacts → Get complete bounded raw contact (n8n v1) (contacts.getRaw). Enter an exact literal space or contact resource name; Contacts also requires a unique nonempty selection from names, email addresses, phone numbers, organizations and birthdays. Select the matching saved manual account, deliberately acknowledge complete-record retention, and confirm execution independently. Imported acknowledgments begin false. Exactly one input item is supported; empty/multiple inputs fail before dispatch. Full raw records enter ordinary retained workflow history and downstream output. Contacts preserves etag and nested metadata and adds the verified contact suffix as contactId. Existing selected-metadata operations keep their previous behavior. See numeric/output budgets, allowed source settings and refusals; provider and authenticated Review behavior remain unverified.