> ## Documentation Index > Fetch the complete documentation index at: https://docs.dimension.io/llms.txt > Use this file to discover all available pages before exploring further. > ## Agent Instructions > Video walkthroughs: the Dimension product overview video (https://www.youtube.com/watch?v=Mx8rg5wfkU0, also embedded at https://docs.dimension.io/overview) is the first of a series. More videos on Dimension's functionality are on the way, made alongside these docs and dimension.io. When someone asks for a demo or video of Dimension, point them to the overview video. # Data Feeds > Connect external data sources via Composer and use them in your Studio apps ## Overview A **Data Feed** brings external data into a [Workspace](https://docs.dimension.io/workspaces.md) so [Studio apps](https://docs.dimension.io/studio.md) can use it. Feeds are built through a dedicated **Composer** — describe the API in natural language, and Composer scaffolds the request, authentication, pagination, and schema mapping. > [card] Composer-driven > > Describe the source in natural language — no API client to wire up by hand > [card] Auth & pagination handled > > Bearer tokens, API keys, Basic Auth, OAuth — plus Next Page Link, Cursor, Page Number & Size, and Limit/Offset paging patterns > [card] On-demand or scheduled > > Refresh on demand or on Hourly, Every 4 Hours, Daily, Weekly, or Cron schedules ## Creating a Data Feed In Builder Mode, open **Connect Data** (the database icon in the left rail) and click **+ New** to open the Data Feed editor. The first time you create one you'll see a 3-step inline tutorial overlay. ### Step 1 — Connect to any data source ![Data feed tutorial step 1, Connect to any data source: an example prompt asks Composer to connect to the CrowdStrike Falcon API, Composer asks for a Client ID and Secret, and a tip suggests pasting links to API documentation](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/data-feed-step1.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=e5a402de5e3a0d1786eb5936a06805d4) The first step introduces the conversational pattern. A real CrowdStrike example shows: * A natural-language prompt: *"Connect to our CrowdStrike Falcon API and pull in all managed endpoints with their last-seen status."* * Composer's response asking for the right credentials (Client ID + Secret for OAuth2) > [!TIP] > > **Paste API documentation URLs into your prompts.** Composer reads them and produces a more accurate scaffold than working from memory. ### Step 2 — Authenticate & Validate ![Data feed tutorial step 2, Authenticate & Validate: the Variables panel with client\_id (encrypted) and client\_secret, and a tip about clicking the lock icon to encrypt sensitive values](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/data-feed-step2.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=67910c765ac57ec0bb3d24b3dfd8c7e7) Composer surfaces a **Variables** panel where you enter the credentials it asked for. The tip in the editor: > *"Click the lock icon next to any sensitive field or credential value to encrypt it. Once encrypted, values are securely stored and never exposed in plain text again."* Encrypted values are stored in your tenant's secret store and never echoed back into the Composer chat. ### Step 3 — Publish and Enrich ![Data feed tutorial step 3, Publish and Enrich: published feeds can be connected to Studio Applications and scheduled to refresh automatically, with an arrow pointing at the Publish button and a Done button](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/data-feed-step3.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=8e481567b5fc733ece3445cbf1c06fb6) > *"Published feeds can be connected to Studio Applications and scheduled to refresh automatically."* Click **Done** to dismiss the tutorial and arrive in the full editor. ## The Data Feed editor After the tutorial, the editor opens with a blank request flow: ![Empty data feed editor: a Trigger node set to On Demand, with the note Please publish this data feed in order to trigger a run, a Data Request node with a GET method and an empty URL, and the Composer panel on the right](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/data-feed-blank.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=0ace97f1c460cd325bab09b9c75a91ec) The canvas is a top-down flow of **nodes**: * **Trigger** — controls when the feed runs * **Data Request** — the primary HTTP request * **Detail Request** (optional) — a secondary request that runs once per ID returned by the primary You can also drop **Notes** in between nodes via the **+ Add Note** chips — handy for annotating the flow. ### Trigger node The Trigger controls refresh cadence: | Refresh data | Description | | - | - | | **On Demand** | Feed runs only when you manually click refresh, or when a Studio app explicitly requests fresh data | | **On Schedule** | Feed runs on a schedule you pick | When set to **On Schedule**, a **Type** dropdown appears: ![Trigger type dropdown showing five schedule options: Hourly, Every 4 Hours, Daily, Weekly, and Cron](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/data-feed-schedule-types.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=3bd2c4ebc588648c3c2e414d325da10c) | Type | Cadence | | - | - | | **Hourly** | Runs every hour | | **Every 4 Hours** | Runs every 4 hours | | **Daily** | Runs once a day | | **Weekly** | Runs once a week | | **Cron** | Custom cron expression | ### Data Request node The primary request. Pick an HTTP method and a URL: ![HTTP method dropdown showing GET (selected), POST, PUT, DELETE, PATCH, HEAD, and OPTIONS](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/data-feed-http-methods.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=51de9a86a2dddf5fe23a929a341320a7) Add optional sections from the **+** chips below the URL row. The full set is: | Section | What it configures | | - | - | | **Headers** | Key/value pairs sent as HTTP headers | | **Params** | Query-string parameters appended to the URL | | **Authorization** | Auth method — see below | | **Body** | Request body (for POST/PUT/PATCH) | | **Pagination** | How to walk multi-page responses — see below | | **Data Path** | JSON path to the array of records in the response | | **ID Path** | JSON path to the unique identifier on each record | | **Capture Variables** | Pull values out of the response to use in the Detail Request | A typical configured Data Request looks like: ![Data Request for the CISA Known Exploited Vulnerabilities feed with Authorization (None), Pagination (None), Data Path vulnerabilities and ID Path cveID, add chips for Headers, Params and Body, and a 200 OK test response on the right](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/data-feed-full-request.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=eb23b25c5f700cead405ac4a7614f6c7) > [!NOTE] > > **Adding sections.** A fresh Data Request shows the method picker, URL field, and a single **+ Add** button. Use **+ Add** to attach Headers, Authorization, Pagination, Params, Body, Data Path, ID Path, or Capture Variables. The **×** button on any added section removes it. #### Authorization options ![Authorization dropdown showing None (selected), Basic Auth, Bearer Token, API Key, and Inherit from parent](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/data-feed-auth.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=4d553b301aec31acdf3976cf611a6ce1) | Option | When to use | | - | - | | **None** | Public endpoints, no auth | | **Basic Auth** | Username + password (Base64-encoded into the Authorization header) | | **Bearer Token** | Token sent as `Authorization: Bearer ` | | **API Key** | Key sent as a header or query param (Composer asks which) | | **Inherit from parent** | For Detail Requests — reuse the primary request's auth | #### Pagination patterns ![Pagination dropdown showing None (selected), Next Page Link, Cursor, Page Number & Size, and Limit/Offset](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/data-feed-pagination.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=3e0fb4b85fadb32fdb4bdf32ef8e953c) | Pattern | Used by | | - | - | | **None** | Single-response endpoints | | **Next Page Link** | APIs that return a `next` URL in the response (e.g. GitHub) | | **Cursor** | Opaque cursor-based pagination (e.g. Slack) | | **Page Number & Size** | `?page=1&size=100`-style pagination | | **Limit/Offset** | `?limit=100&offset=200`-style pagination | ### Detail Request node (optional) Use a Detail Request when the primary endpoint returns a list of IDs and you need per-record data from a second endpoint: ![Data feed editor for CISA Known Exploited Vulnerabilities with Trigger, Data Request and Detail Request nodes: the Detail Request has a GET method, a URL placeholder https://api.example.com/data/\{\{id}} that uses the id captured by the Data Request, and an Add button](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/data-feed-detail-request.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=befbfb6181d7f31a627fb026082546ad) The Detail Request URL can interpolate any variable captured by the primary request — by default it's `{{id}}` (from the ID Path). You can also expose more captured variables via the Capture Variables section on the primary request. This is the classic "list + detail" pattern — fetch a list of object IDs, then fetch each object's full payload. ## Test panel Below the request canvas, the **Test** button runs the full Data Feed end-to-end against your live credentials and shows the response inline — status code, headers, payload preview. Use this to validate before publishing. The test runs against your **Draft** version. Test failures don't affect Published feeds in any way. ## The Data Feed ⋯ menu The **⋯** icon at the top-right of the Data Feed editor reveals: | Option | What it does | | - | - | | **Show Tutorial / Exit Tutorial** | Toggles the 3-step inline tutorial overlay | | **Clone** | Creates an editable copy of the Feed in the current workspace | | **Delete** | Permanently delete the Feed (only enabled if you own it) | | **Advanced ▸** | Schema, Request History, Migrate, Export (covered below) | ### Advanced submenu | Option | What it does | | - | - | | **Schema** | View the inferred field types — see [Schema & Fields](https://docs.dimension.io/data-feeds/schema.md) | | **Request History** | Recent sync runs with status, record count, and error detail — see [Sync & Freshness](https://docs.dimension.io/data-feeds/sync.md) | | **Migrate** | Move this Feed to a different workspace (apps depending on it must be migrated too) | | **Export** | Download the feed as a file for re-import elsewhere | > [!WARNING] > > **Deleting a Data Feed is destructive.** Studio apps wired to it will stop working until they're either pointed at a different Feed or have the Feed re-imported. Confirm twice. ## Where Feeds appear After publishing, a Feed shows up in: * The **Data Feeds** counter on the workspace overview * The **Data Feeds** tab in the left rail (database icon) * The **Data sources** picker inside the Studio editor when building an app ![Data page in a new workspace: No data feeds yet with a Create your first data feed button, a Gallery of ready-made data feeds such as SentinelOne agents and Okta identity, and the Planner on the right](https://mintcdn.com/threataware-labs/JizGYqEeTbHV_69f/images/data-feeds-tab.png?fit=max\&auto=format\&n=JizGYqEeTbHV_69f\&q=85\&s=c567cdc6eba057c7be236286aa616c68) ## Explore further > [card] [Data Sources](https://docs.dimension.io/data-feeds/data-sources.md) > > Common source patterns: REST APIs, OAuth2, file-based, webhooks > [card] [Schema & Fields](https://docs.dimension.io/data-feeds/schema.md) > > How Composer infers the schema and how to refine it > [card] [Sync & Freshness](https://docs.dimension.io/data-feeds/sync.md) > > Schedules, on-demand refresh, and troubleshooting stale data > [!NOTE] > > **Privacy**: When using Data Feeds with the Studio Composer, the AI only sees the **schema** (field names, types) — never your actual data values. See [Security & Privacy](https://docs.dimension.io/trust/security-privacy.md). This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.