> ## Documentation Index
> Fetch the complete documentation index at: https://docs.deepmerge.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Just-in-time context retrieval

> What your AI tools can do with get_context.

<Note>Action name: `get_context` · Read only (never changes anything)</Note>

Load just-in-time context for a task. The default retrieval entry point:
pass what you are about to do and the workspace returns the most relevant
memory pages (the consolidated brain), ranked by keyword + semantic
similarity.

## Inputs

| Field   | Type    | Required | What it is                                                                          |
| ------- | ------- | -------- | ----------------------------------------------------------------------------------- |
| `task`  | string  | Yes      | Required. What you are about to do, in natural language. Used as the ranking query. |
| `limit` | integer | No       | Max pages returned. 1..20, default 8.                                               |

<Accordion title="Full instructions the AI sees">
  ```text theme={null}
  CALL THIS BEFORE STARTING ANY NON-TRIVIAL TASK. Skipping it means you
  rebuild context the team already established.

  USE WHEN: starting any non-trivial work; loading context after a fresh
  session; deciding whether to redo work the team already did.

  Page bodies are returned as previews (capped); pass an id to `fetch` for
  the full document. Returns { task, memories }.
  ```
</Accordion>