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

# Connect Context over MCP

> Use Context from Claude, Cursor and other MCP clients.

Context's MCP server is a remote server at:

```text theme={null}
https://jabrod.com/api/mcp/context
```

It speaks Streamable HTTP and answers each request with JSON. Tools act on the workspaces of the account you signed in with, or that owns the API key.

## Authentication

There are two ways in. Both send `Authorization: Bearer …` with every request.

### Sign in with Jabrod

Add the server without a key. The first time your client connects, it opens Jabrod's sign-in page, auth.jabrod.com, in your browser: sign in, and allow the app to use your account. The token it gets works only for this server.

In Claude Code:

```bash theme={null}
claude mcp add --transport http jabrod-context https://jabrod.com/api/mcp/context
```

Then run `/mcp` in Claude Code, choose `jabrod-context`, and sign in.

This needs a client that supports MCP authorization. [Sign in with Jabrod](https://docs.jabrod.com/platform/sign-in.md#mcp-clients) explains the flow.

### API key

For clients that can't sign in, make a personal API key at [jabrod.com/app/keys](https://jabrod.com/app/keys). It starts with `jbd_sk_` and is shown once. Send it with every request:

```http theme={null}
Authorization: Bearer jbd_sk_…
```

Make one key per app, so you can revoke one without breaking the others. A revoked key stops working at once.

## Client configuration with an API key

> [tab] Claude Code
>
> ```bash theme={null}
> claude mcp add --transport http jabrod-context https://jabrod.com/api/mcp/context --header "Authorization: Bearer jbd_sk_…"
> ```

> [tab] Cursor
>
> ```json .cursor/mcp.json theme={null}
> {
>   "mcpServers": {
>     "jabrod-context": {
>       "url": "https://jabrod.com/api/mcp/context",
>       "headers": { "Authorization": "Bearer jbd_sk_…" }
>     }
>   }
> }
> ```

> [tab] Claude Desktop
>
> Claude Desktop's config file starts local programs, so connect through `mcp-remote`, which forwards to the remote server. In **Settings ▸ Developer ▸ Edit Config**:
>
> ```json claude_desktop_config.json theme={null}
> {
>   "mcpServers": {
>     "jabrod-context": {
>       "command": "npx",
>       "args": ["-y", "mcp-remote", "https://jabrod.com/api/mcp/context", "--header", "Authorization:${JABROD_KEY}"],
>       "env": { "JABROD_KEY": "Bearer jbd_sk_…" }
>     }
>   }
> }
> ```

> [tab] Other clients
>
> Any client that supports Streamable HTTP servers with custom headers works. URL `https://jabrod.com/api/mcp/context`, header `Authorization: Bearer jbd_sk_…`.

## Tools

| Tool | Input | Returns |
| - | - | - |
| `list_workspaces` | none | `[{id, name, documents, ready, created_at}]` |
| `create_workspace` | `name` | `{id, name}` |
| `list_documents` | `workspace_id` | `[{id, title, kind, status, chunks, error}]` |
| `add_document` | `workspace_id`, and either `url`, or `title` and `text` (Markdown or plain text, up to 2 MB) | `{id, status: "queued"}` |
| `delete_document` | `document_id` | `{deleted: true}` |
| `search` | `workspace_id`, `query`, `k?` (1 to 20, default 6) | `[{document_id, title, locator, text}]`, best first |
| `ask` | `workspace_id`, `question` | `{answer, refused, citations: [{n, document_id, title, locator, text}]}` |

### Notes

* **Adding sources is asynchronous.** `add_document` returns at once with `status: "queued"`. Call `list_documents` until the source is `ready` or `failed`.
* **`search` writes no answer.** It returns the passages [retrieval](https://docs.jabrod.com/context/cited-answers.md#retrieval) finds, so your own model can use them.
* **`ask` can refuse.** When the sources don't cover the question, `refused` is `true` and the answer says so. See [Refusals](https://docs.jabrod.com/context/cited-answers.md#refusals).
* Questions asked over MCP are saved with the workspace, but are not offered as [Evals](https://docs.jabrod.com/evals/test-sets.md) test cases.

## Errors

A problem you can fix comes back as a tool result with `isError: true` and a sentence saying what to do, such as "That workspace wasn't found." Requests without a valid key get HTTP 401.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.