autario in n8n
n8n-nodes-autario is a community node for n8n, the open workflow automation tool. It wraps the autario REST API so a workflow reads or writes a dataset the way it talks to a database, without an HTTP Request node per call. The node is declarative, which means every operation is a described request against a documented path rather than custom code, and it has no runtime dependencies. This page is the install guide, a step-by-step tutorial for a first workflow, the setup for using autario as a tool of an n8n AI Agent, the list of operations, troubleshooting and the honest account of what the node does not do.
Install the node
On a self-hosted n8n the node installs from the interface. Community nodes are a self-hosted feature; on n8n Cloud the node becomes available once n8n has verified it. We have submitted the node for verification and it is in review, which n8n says takes up to four weeks and which we do not control.
- In n8n open Settings, then Community nodes, then click Install.
- Enter the package name n8n-nodes-autario, tick the acknowledgement and click Install.
- Open a workflow, click the plus to add a node and search for Autario. It appears with the autario icon and three resources: Dataset, Row and Connector.
npm install n8n-nodes-autario
The command above is the same package for an image-based install, where the operator adds community packages at build time. The package is MIT licensed and its source is public; both links are at the end of this page.
Connect your account
The node uses one credential, Autario API, with two fields: the API key and the API secret. Both are created at https://autario.com/account under API keys, and the secret is shown once. n8n sends them as the x-api-key and x-api-secret headers on every call, the same two headers the REST API documents for every client.
The credential test calls the listing of your own datasets. It answers your tables and nothing else, so a passing test proves the pair is valid without touching the public catalog. A key with a preset or app scope works the same way: the node can only reach what the key can reach.
GET https://autario.com/api/v1/me/datasets
x-api-key: <your key>
x-api-secret: <your secret>
Tutorial: your first autario workflow
This tutorial builds one workflow that finds public datasets, creates a dataset of your own, writes two rows into it and lists your connectors. It needs a self-hosted n8n with the node installed and an autario account; a free account works. Every click below names the label n8n shows.
- In autario open https://autario.com/account, go to API keys and click Create API key. Copy the key and the secret; the secret is shown once.
- In n8n create a new workflow and add the first step Trigger manually, so the workflow runs when you click Execute workflow.
- Click the plus after the trigger, search for Autario and pick the action Search datasets.
- In Credential to connect with choose Create new credential. Paste the API key and the API secret into the Autario API credential and click Save. n8n tests the pair right away against the listing of your own datasets, so a green test means the key works.
- Set Search Term to a topic, for example unemployment, and click Execute step. The output is one item holding the matching datasets from your own tables, those shared with you and the public catalog, each with its ID.
- Add the next node: Autario, action Create a dataset. Give it a Title such as n8n tutorial and click Execute step. The output carries the id of the new, empty dataset.
- Add Autario, action Append rows to a dataset. In Dataset ID switch to Expression and enter {{ $('Create a dataset').item.json.id }}, which reads the ID from the step before. In Rows enter a JSON array of row objects: [{"country":"DE","value":42},{"country":"FR","value":37}]. The keys of the first rows become the columns of the dataset.
- Add Autario, action Get many connectors, as the last step. It returns one item per connector instance with status, schedule and tables.
- Click Execute workflow. Then open My Data on autario.com: the dataset n8n tutorial is there with its two rows, and from there it can be charted, shared or read by an app.
Tip on multi-item steps: an n8n node runs once for every item it receives. Get many connectors returns one item per instance, so a Create a dataset placed after it would create one dataset per connector. Open that node, go to Settings and turn on Execute Once, and it runs a single time whatever arrives. The same switch fits any write step that follows a list.
Tip on empty results: a node that returns no items ends its branch, and the nodes after it do not run. An account without connectors gets no items from Get many connectors, which is why this tutorial puts that step last.
Use autario as a tool for an n8n AI Agent
The node is marked as usable as a tool, so an n8n AI Agent can call it on its own: the model decides when to search autario and with which term, n8n runs the request, and the answer goes back into the conversation. The model runs on your own key with your own model provider; autario receives only the REST calls of the tool and pays for no model call.
- Create a new workflow, add the chat trigger (search for Chat in the trigger list) and connect an AI Agent node to it.
- Under the AI Agent click the plus at Chat Model and pick the model node of your provider, for example OpenAI Chat Model or Anthropic Chat Model. Create its credential with your own API key from that provider and choose a model.
- Click the plus at Tool, search for Autario and choose Search datasets. Select the Autario API credential from the tutorial above.
- Hover the Search Term field and click the button Let the model define this parameter. The field now shows that the model fills it at run time instead of a fixed value.
- Close the tool, click Open chat at the bottom of the canvas and ask: Which autario datasets cover youth unemployment in Europe? List titles and IDs.
- The agent calls Search datasets with a term it picked, reads the result and answers with titles and IDs. The execution log in n8n shows each tool call with its input and output.
More tools attach the same way. Add Autario with Get the schema of a dataset and leave Dataset ID to the model, and the agent can check columns before it answers; add Get many datasets you own, and it can answer questions about your own tables. Each tool is one operation, so the agent only gets the operations you attach.
n8n 2.x also has a separate Agent Builder. The autario tool described here attaches to the AI Agent node inside a workflow, which is the path this page documents.
What the node can do
Every operation is one documented path of the autario OpenAPI document, and the node description names the path next to each operation, so there is nothing to look up twice. The Dataset resource reads, the Row resource writes, the Connector resource lists and refreshes.
- Dataset, Search: the datasets your key can read, your own, those shared with you and the public catalog, by a search term with a limit and an offset. GET /datasets.
- Dataset, Get and Get Schema: the metadata of one dataset, and its columns with types, row count and one semantics sentence per column. GET /datasets/{id} and GET /datasets/{id}/schema.
- Dataset, Query: filtered, sorted, paginated rows. Filters are written one per line as column:op:value with the operators eq, neq, gt, lt, gte, lte and like, and travel as the repeatable filters parameter. Fields, sort, limit, offset and the overlay mode follow the API. GET /datasets/{id}/query.
- Dataset, Get Many: every dataset you own, with a query URL and a CSV URL per row, optionally narrowed to one brand or some connector instances. GET /me/datasets.
- Dataset, Export CSV: the whole table as CSV text with no paging, returned in a csv field; an empty table answers an empty string. GET /datasets/{id}/export.
- Dataset, Create: an empty dataset you own, whose schema comes from the first rows appended. POST /datasets.
- Row, Append: a JSON array of row objects into a dataset you own, at most ten thousand rows per call. POST /datasets/{id}/rows.
- Row, Clear: every row goes, the schema and the dataset ID stay. DELETE /datasets/{id}/rows.
- Connector, Get Many and Get: your connector instances with status, schedule, tables and health, one item per instance. GET /connectors and GET /connectors/{id}.
- Connector, Refresh: pull the vendor API now and rewrite the hosted tables. POST /connectors/{id}/refresh.
What the API refuses, the node passes through unchanged. A dataset that is not yours answers 404, which is deliberately the same answer a dataset that does not exist gives. A dataset an app owns answers 409 app_owned_dataset on Append, because those rows are checked against the declared columns of the app inside the app. A connector refresh less than ten minutes after the last run answers 429 with the time the next run may start, and a connector whose sign-in is gone answers 409 with the URL where the owner signs in again.
A workflow that keeps a table fresh
The pattern most workflows end up with is a scheduled full replace: something produces rows, autario stores them under a stable ID, and everything downstream on autario, charts, apps and shares, keeps reading the same dataset.
- A Schedule Trigger fires once a day.
- A node produces the rows: a database query, a spreadsheet, an HTTP call to a system autario has no connector for.
- Autario, Row, Clear on your dataset, so the table starts empty but keeps its ID.
- Autario, Row, Append with the rows as a JSON array. Rows above ten thousand go in batches with a Split In Batches node.
- Autario, Dataset, Query to read the fresh table back and post a summary wherever the team reads.
The other direction works the same way. A Connector, Refresh at the start of a workflow brings the vendor tables up to date, and a Dataset, Query or Export CSV afterwards hands the rows to the rest of the workflow, for example into a report that n8n sends by mail.
What it does not do
- It does not create connectors. A connector means a credential for a vendor API, and that is set up by the account owner in the autario interface, never by a workflow.
- It does not call the MCP server. As an AI Agent tool it runs the same REST operations listed above; an agent that should reach the full set of autario MCP tools connects to the remote MCP server instead, described in the MCP tool reference.
- It does not page for you. Query returns one page with the total in its meta block; a workflow that needs everything uses Export CSV, which has no row cap.
- The statistics endpoints, charts, apps and overlays are not operations of this version. They remain available over REST and MCP.
Every read of your own data through the node is recorded in your access log at https://autario.com/account under Access, with the credential and the row count, the same as any other API client.
Troubleshooting
- Install fails with spawn npm ENOENT. This happens on Windows when n8n was started with npx n8n: n8n starts npm to install a community node, and on Windows the command is npm.cmd, which that start path does not find. Run n8n in Docker instead, where the install from Settings works; the command is below.
- The credential test fails. The key or the secret is wrong, or the key was deleted. The secret is shown only once, so if it is lost, create a new API key at https://autario.com/account under API keys and paste both values again.
- Append rows to a dataset answers Bad request. The Rows field is empty or is not a JSON array. It needs square brackets around one or more row objects, for example [{"country":"DE","value":42}], or an expression that returns such an array.
- Create a dataset made several datasets. The node received several items and ran once per item. Turn on Settings, Execute Once on that node.
- Dataset ID shows no value. The expression {{ $('Create a dataset').item.json.id }} reads the output of that step, so it stays empty until Create a dataset has run in the same execution. Run the workflow from the start with Execute workflow, and keep the node name in the expression identical to the name of the node on the canvas.
- A dataset answers 404. The ID belongs to a dataset your key cannot write or does not exist; writes only reach datasets you own.
docker run -it --rm -p 5678:5678 -v n8n_data:/home/node/.n8n docker.n8n.io/n8nio/n8n
The command starts n8n on http://localhost:5678 and keeps its data, including installed community nodes and credentials, in the Docker volume n8n_data, so a restart keeps the autario node.
Frequently asked questions
How do I connect autario to n8n? Install the community node n8n-nodes-autario under Settings, Community nodes, then create the Autario API credential with an API key and secret from https://autario.com/account. Every Autario action in a workflow then uses that credential.
Can an n8n AI Agent use autario as a tool? Yes. Attach Autario under Tool of an AI Agent node, pick an operation such as Search datasets and let the model define its parameters. The agent decides when to call it.
Does the node work on n8n Cloud? Not yet. Community nodes on n8n Cloud must be verified by n8n first; the autario node is submitted and in review. Until then it runs on any self-hosted n8n, including the Docker image.
Does autario see or pay for my model calls? No. The Chat Model of the AI Agent runs on your own provider key. autario receives only the REST requests the tool makes, and those count like any other API call of your plan, see Limits and pricing.
Which data can a workflow read and write? It reads what the API key can read: your own datasets, those shared with you and the public catalog. It writes only to datasets you own, and a key with a preset or app scope narrows both.
Does the node cost anything? The package is free and MIT licensed. The calls it makes are ordinary autario API calls under the limits of your plan.
Source, package and related pages