# Run an existing workflow as a tool and get its result

> Run an existing workflow as a tool and get its result: 0 companies fully do it, 13 with a limit, 54 carry the tag. Every listing in Workflow automation with the reason and a source.

- Canonical: https://www.anchorterminal.com/niches/workflow-automation.run-workflow-as-tool/
- Markdown: https://www.anchorterminal.com/niches/workflow-automation.run-workflow-as-tool/index.md (~1,800 tokens)
- Slim: https://www.anchorterminal.com/niches/workflow-automation.run-workflow-as-tool/index.min.md (~380 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/niches/workflow-automation.run-workflow-as-tool/index.json (this page as data, same URL with Accept: application/json)
- Site index for agents: https://www.anchorterminal.com/llms.txt (the key docs and every listing in one file, at most 16 MB, about 4 million tokens: https://www.anchorterminal.com/llms-full.txt)
- API: https://www.anchorterminal.com/api/v1/index.json
- MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key)
- DNS: dig +short TXT <slug>.grade.anchorterminal.com (the grade as a DNS TXT record, for a sandbox with no HTTP where names still resolve; also top., best.<category>. and <a>.vs.<b>. under the zone, and the same under grade.letme.dev)
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)
- Updated: 2026-11-15

No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.

| Companies, full fit | With a limit | Carry the tag | Share of the category's work | First-hand accounts |
| --- | --- | --- | --- | --- |
| 0 | 13 | 54 | 15 | 9 |

## What counts as done

The published (not test) version runs once with the given input, the run appears in the platform's log, and the output or a pollable run ID returns within the client's tool timeout; a failed run comes back as an error, not a 200 with error text.

How agents ask for it: run the weekly report scenario; trigger the onboarding flow for this customer; call my n8n workflow with this input; run the enrich-lead script

## Does the job, with a limit

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

- **Activepieces API + MCP (Activepieces Inc.)**: The MCP tool list has test tools but no tool that runs a published flow; acc-14 reports ChatGPT test-running flows with no logged run. Sync webhooks run the published flow but close at 30 s, and issue 13334 reports silent 200/204 on timeout.
- **Microsoft Copilot Studio MCP (Microsoft Corporation)**: The only run path is manage_workflows' real test of a published, enabled workflow, read back through list_runs and get_run; forwarded calls get a 36-minute deadline. It's a build bridge, not a tool surface for running named workflows, and each run costs credits.
- **Gumloop (AgentHub Inc. (doing business as Gumloop))**: POST /agents/{id}/sessions returns 202 and a session to poll, with a caller session_id that 409s on repeat. Webhook triggers return success at once without output. There is no published-versus-test choice, and API access needs Pro.
- **Kestra (Kestra Technologies SAS)**: McpToolTrigger registers a flow as a named MCP tool with inputs as schema and outputs returned, and executions take wait=true and kind=NORMAL. No MCP timeout is stated, and the docs don't show a failed execution returning an MCP error.
- **Kognitos (Kognitos, Inc.)**: kognitos_invoke_automation takes stage PUBLISHED and an idempotency_key and returns a run to poll with kognitos_runs. A run can stop in waiting for a person, and failure arrives as run state on poll; no MCP error example is documented.
- **Make API + MCP (Celonis, Inc.)**: MCP runs active and on-demand scenarios as tools with inputs and outputs; a call past 25 s (OAuth) or 40 s (key) returns an executionId to fetch later, and runs last up to 40 minutes. How a failed run reaches the client isn't documented.
- **n8n API + MCP (n8n GmbH)**: execute_workflow takes executionMode production and returns an execution ID at once, polled with get_workflow_execution. An MCP Server Trigger failure came back without isError (acc-09), and a production webhook failed with no execution logged (acc-36).
- **Paragon ActionKit + MCP (Forge Technology, Inc. (d/b/a Paragon))**: Request triggers and ActionKit's POST /tools run for one Connected User, and the docs' AI SDK example treats non-2xx as failure, but the listing records Run Tool returning 200 and dropping unknown parameters, and no run ID is described for ActionKit calls.
- **Pipedream API + MCP (Pipedream, LLC)**: Existing workflows keep running on their triggers until the shutdown, and Connect runs single prebuilt actions as MCP tools with isError on failure, but a workflow itself isn't exposed as a tool and the product is ending.
- **Microsoft Power Automate (Microsoft Corporation)**: The page documents no run call; an agent starts a flow only through an HTTP request trigger the owner added, and the run lands in the FlowRun table. Whether a failed run returns an error to the caller wasn't documented in what was read.
- **Prismatic (Prismatic Software Inc.)**: The hosted MCP flow server lists agentic flows as tools with invocation schemas, scoped globally, per integration or per instance. Organisation members see only test instances; customer users see production ones. No timeout or failure shape is documented on the pages read.
- **Tray.ai API + MCP (Tray.ai, Inc.)**: Agent Gateway exposes a workflow with an Agent Tool Trigger and Trigger Reply as one MCP tool returning structured results. Agent Gateway is an add-on, no timeout is stated, and the listing records upstream errors wrapped in a 200.
- **Windmill API + MCP (Windmill Labs, Inc.)**: Scripts and flows as tools gives one typed tool per selected path, and the call waits for the result, but a wait that times out cancels the run unless the flow uses early return. acc-18 hit a separate 1,000-execution MCP cap (still open).
- **Workato API + MCP (Workato, Inc.)**: Recipes can be turned into skills and MCP servers, and force_run starts an on-demand run, but the pages read don't say whether a tool call waits for the job or how a failed job reaches the caller.

- not this job: Bind AI (1Fort Inc.): No MCP server or run endpoint for a named workflow; the API is the web app's backend behind a user JWT from sign-in or OTP.
- not this job: Cooper (Mandl) (Cooper AI Tech, Inc.): No run endpoint, webhook or MCP tool is published.
- not this job: Motive Agents (Motive Technologies, Inc.): Agents fire on Motive's own safety, maintenance and idling events; nothing lets an outside agent call one.
- not this job: Node-RED (OpenJS Foundation): No run endpoint and no execution log; an agent can only call an HTTP In node the owner built, and nothing records that the run happened.

## What stops agents finishing

- tool call times out before the run finishes and no run ID is returned
- client test-runs the flow instead of running the published version
- separate MCP execution cap below the plan's run allowance
- tool schema rejected by one model family (Gemini)
- production webhook fails without logging an execution

## Around it

Before: workflow-automation.build-workflow, chat.user-asks. After: email.send, crm.update-record, workflow-automation.debug-failed-run.

## Where this comes from

Weight 15 of 100 in Workflow automation: 9 of 48 accounts (n8n, Activepieces, Windmill, Kestra); MCP and webhook run failures in GitHub issues and the n8n forum

As data: https://www.anchorterminal.com/api/v1/use-cases/workflow-automation.run-workflow-as-tool.json, or the MCP tool get_use_case with id workflow-automation.run-workflow-as-tool. Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person.

---

Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp
