Ground your agents.
Connect with MCP.

Connect your MCP client directly to DocSlurp over Streamable HTTP. Discover tools and search your workspace documents without installing a bridge.

Connect your assistantStreamable HTTP · OAuth

In Claude, open Customize → Connectors → Add custom connector. On a team, an organization owner may need to add it first.
Name it DocSlurp and paste this URL. Choose browser sign-in and “Register automatically” for the OAuth client, then connect.
Enable the connector for your conversation. The remote connector also works in Claude Desktop.

Remote MCP server URL

https://docslurp.io/v1/mcp

Claude connector guide ↗Verified October 2026

API key configuration / self-hosted endpoint

Connect an MCP client

Give your agent access to workspace search through our hosted MCP server. Sign in with OAuth or use a server API key.

Streamable HTTP endpoint

https://docslurp.io/v1/mcp

MCP clientAuthentication

Add this to .vscode/mcp.json. Start the server in your client, then sign in and approve access in your browser. No API key is needed.

.vscode/mcp.json

{
  "servers": {
    "docslurp": {
      "type": "http",
      "url": "https://docslurp.io/v1/mcp"
    }
  }
}

MCP setup and authentication docs →

OAuth sign-in

Add the server URL to your client and connect. Your client discovers the authorization server, opens DocSlurp in your browser, and asks you to approve access. Sign in with your normal DocSlurp account. Authorization uses a code flow with PKCE; your client handles the token exchange.

API key authentication

For unattended agents, create a server key with search scope in API keys. Choose API key above and replace the placeholder. Authentication uses an Authorization: Bearer header. Client keys intended for browser widgets do not grant MCP access. Keep your completed config out of source control; revoke the key when it is no longer needed.

Tools and access

Your client discovers available tools through tools/list. Workspace search requires a workspace ID and a query, and returns results with source citations. Access is checked against the authenticated organization and credential permissions on every request.

Tool	Use it for
`workspace.search`	Retrieve text and citations to ground any LLM. No answer generation.
`workspace.search-many`	Run up to 20 queries with at most four in flight. Results stay in input order, including per-query errors.
`workspace.chat`	Get a grounded answer, citations, session ID and token usage. Uses the workspace's configured model.

Start with { "workspaceId": "YOUR_WORKSPACE_ID", "q": "What are the support hours?" } for search or chat. Pass the returned sessionId to chat when continuing a conversation. For multiple queries, use { "queries": [{ "workspaceId": "YOUR_WORKSPACE_ID", "q": "Support hours?" }, { "workspaceId": "YOUR_WORKSPACE_ID", "q": "Escalation process?" }] }.

Search returns portable passages and source metadata that you can send to your own model. Retrieval can incur embedding usage; chat additionally incurs generation usage. Search responses include a usage object with inputTokens, outputTokens, totalTokens, costCents (USD), and costSource. Batch search includes usage per query. These totals cover retrieval AI only; costSource identifies provider-reported versus estimated costs. Chat usage covers answer generation. Search, parallel retrieval and chat require search scope.

The remote server uses Streamable HTTP at /v1/mcp. This is the MCP protocol endpoint, rather than the older REST tool-call paths. Clients can make normal HTTP requests without maintaining a long-lived connection.

Optional local bridge

Clients that only support stdio can still use docslurp mcp-server from @docslurp/cli. Set DOCSLURP_API_KEY in the bridge process environment; DOCSLURP_URL optionally selects a local or self-hosted service.

Configuration follows the VS Code MCP reference and Cursor MCP documentation. Use a current client version with Streamable HTTP and OAuth support.