REST API Connector
A generic HTTP integration. The connection captures the base URL and authentication scheme;
each Data Source node that uses it carries its own AI-generated fetch function describing
exactly which endpoints to hit and how to paginate.
What it lets you do
- Save a base URL and credentials once, reuse it across many Data Source nodes.
- For any node, describe in plain language what data you want; the AI generates a custom fetch function that handles authentication, pagination, and shaping the response into rows.
- Use four authentication schemes out of the box.
Connection details required
| Field |
Sensitive |
Default |
| Base URL |
No |
(empty) |
| Auth type |
No |
None |
| API key header name |
No |
X-API-Key (only for API-key auth) |
| Bearer token |
Yes |
(only for bearer auth) |
| API key value |
Yes |
(only for API-key auth) |
| Basic username |
No |
(only for basic auth) |
| Basic password |
Yes |
(only for basic auth) |
The form shows only the fields relevant to the auth type you picked.
How a fetch function is built
When you create a Data Source node pointing at this connection and describe what data you want,
the AI generates an asynchronous JavaScript function that:
- Receives an HTTP client.
- Loops over pages, cursor tokens, or related endpoints as the API requires.
- Returns rows as an array of plain objects.
Security
The function uses placeholder tokens ({{BASE_URL}}, {{AUTH_HEADER_NAME}}, etc.)
— never literal credentials — when referencing the connection. The placeholders are filled in with real
values only inside the background engine, just before the function runs. The interactive layer therefore
never holds a decrypted key, and saved workflows contain no secrets even if shared.
Behaviour notes
- The connection itself doesn't perform any request beyond the test-connection probe. Real fetches happen from inside the Data Source node's fetch function.
- A safety cap on the maximum number of pages prevents an unbounded API from looping forever.
- Sensible defaults are baked into generated functions: GET method, JSON Accept header, a request timeout, and a reasonable page size.
- If the API returns nothing, DagFlux still creates an empty collection so downstream nodes have a defined source.
- If the API returns a single object, DagFlux wraps it as a single-element array.
