Ask questions of your connected business data from n8n — and get answers you can check.
DataGrout plans the work from your question, runs it against the systems you have connected, verifies the result against what you actually asked, and returns a certificate showing exactly what ran. When it cannot verify a result it says so rather than answering anyway.
Two nodes:
| Node | Use it to |
|---|---|
| DataGrout | Ask questions, compute over data, store and recall facts, re-run verified work. Also works on an AI Agent's Tool connector. |
| DataGrout Trigger | Start a workflow when a DataGrout run or background task finishes or fails. |
The DataGrout node calls DataGrout's REST API. No runtime dependencies.
- A verified answer, or an admission. Every answer says whether DataGrout's
verifier accepted it, with a certificate showing the steps that produced it.
An answer it could not verify arrives as
verified: false, withanswerempty and what the plan found underunverified_answer— never as a confident answer — so a workflow can branch on it instead of trusting it. - Memory that outlives the execution. Facts stored in one run are still there in the next, and in other workflows pointed at the same memory.
- Compute that stays next to the data. A large result never has to travel through your workflow to be grouped or aggregated.
- Events pushed to you. One WebSocket, no polling, no webhook URL to paste into a dashboard.
| Resource | Operation | What it does |
|---|---|---|
| Answer | Ask | Ask in plain language. DataGrout plans, runs and verifies, and returns the answer plus a certificate URL. |
| Data | Transform | Describe the result you want; DataGrout computes it on its own servers and returns records, one item per row. |
| Memory | Recall / Remember | Store facts as workflows run, then ask what is known — including what can be inferred from it. |
| Skill | Run | Re-run work DataGrout has already verified — a saved skill (skill_…), or the Skill ID an Ask returned. No planning, so it is fast and repeatable. |
| Field | |
|---|---|
answer |
The answer, or empty when it could not be verified |
verified |
Whether DataGrout's verifier accepted it |
unverified_answer, caveat |
What the plan found, and why it could not be verified |
certificate |
id and url of the certificate showing exactly what ran |
skill_id |
Run this verified work again with Skill → Run. An id beginning skill_ does not expire; when skill_expires is true it lasts an hour. |
usage |
Credits charged and remaining |
A large result never has to pass through your workflow. Ask an earlier step for a reference instead of rows, hand that reference to Transform, and the grouping or aggregation happens next to the data — you get back only the answer.
Facts stored with Remember are still there on the next execution, and in other workflows pointed at the same memory. Recall answers from stored facts and from what follows logically from them, with no model call.
DataGrout publishes every server event to one topic, so the trigger subscribes once and you pick the events you want:
| Event | Carries |
|---|---|
| Run Completed | run_id, execution_id, status, tool_name, source, duration_ms |
| Task Completed | task_id, tool_name, cache_ref, status |
| Task Failed | task_id, tool_name, error, status |
| Tool Call Failed | run_id, execution_id, tool_name, error |
Run Completed fires on every terminal status, so failed, timed-out and
cancelled runs arrive there too — the status comes with the event. Leave the
selection empty to receive everything, including events added to DataGrout after
this release.
Task Failed is the one nothing else reports: a background task's caller has
long stopped waiting by the time it fails.
Task Completed carries the result's cache_ref, so the next node can transform
that result server-side without re-fetching or recomputing it.
run_id is the integer DataGrout's own runs.get accepts, so a following node
can fetch the full run without translating anything.
Connection lifecycle frames are filtered out, so a reconnect does not start your workflow. The connection is held open with a keepalive, and a genuinely dropped one reopens on a backing-off delay.
Press Test step in the editor to emit an example carrying the real fields of whichever event you selected, so you can wire downstream nodes before a real event happens.
Import any of these from Workflows → Import from File, then attach your DataGrout credential to the DataGrout nodes.
| File | Shows |
|---|---|
examples/datagrout-ai-agent.json |
A chat AI Agent with DataGrout on its Tool connector. The recommended starting point. |
examples/datagrout-trigger-task-result.json |
A finished background task's cache_ref handed straight to Transform. |
examples/datagrout-trigger-failures.json |
One subscription, branching on event to route failures. |
Settings → Community Nodes → Install → @datagrout/n8n-nodes-datagrout.
To use the node on an AI Agent's Tool connector on a self-hosted instance, set
N8N_COMMUNITY_PACKAGES_ALLOW_TOOL_USAGE=true and restart n8n.
Choose either on the node's Authentication field.
DataGrout API (simplest)
| Field | Where it comes from |
|---|---|
| API Token | Your DataGrout dashboard |
| Server ID | Your DataGrout server UUID |
| Gateway Base URL | Leave as-is unless you self-host the gateway |
DataGrout OAuth2 API — for instances that prefer an OAuth flow. Create the credential, click Connect my account, and there is nothing to fill in: the token itself names the server it was granted.
Testing a credential only reads — it fetches the server's name, changes nothing and costs nothing.
The trigger listens over DataGrout's WebSocket transport, which a new server has switched off. It switches it on when the trigger starts, and writes that to the n8n log, so a change to your server's settings is never silent.
The free-text fields default to $fromAI(...), so an agent can fill them in with
no extra setup — connect the node to the agent's Tool connector and ask a
question. Replace a field with a fixed value to pin it.
- Wait for Result — DataGrout moves slow work to the background; the node waits this long for the finished result. Set 0 to return immediately with the task to follow.
On Transform, Split Rows Into Items (on by default) outputs one item per row of the result. Turn it off to get the whole response as one item, with its certificate and reference.
On the trigger:
- Include Event Name — add the event name to each item as
event. - Reconnect Automatically — reopen a dropped connection.
- Topic — change only to follow a single orchestration run, which publishes to a topic of its own.
Tested against n8n 2.36 on Node 24.
The trigger uses Node's built-in WebSocket, which keeps this package free of runtime dependencies and needs Node 22 or newer. n8n 2.x requires Node 24, so that is already satisfied; on a much older n8n running Node 20 the trigger reports what it needs instead of failing obscurely.
npm install
npm test # unit tests, no dependencies
npm run build
npm run lint