1. Home
  2. Grounded Docs MCP Server
  3. Grounded Docs MCP Server for Cursor: Setup & Config

Data as of

Grounded Docs MCP Server for Cursor: Setup & Config

Cursor: Grounded Docs MCP Server builds a documentation index that coding assistants can query by library version.

What you can do with Grounded Docs in Cursor

Choose it for a version-aware documentation index shared by scripts, a Web UI, and MCP assistants.

  • Fetch a documentation page as Markdown for an agent or script.
  • Index hash-routed documentation sites with explicit route preservation.

Before you start: Grounded Docs in Cursor

  • Node.js 22 or later
  • An MCP-compatible client for MCP access

The README starts this long-running server, opens its Web UI at http://localhost:6280, and connects MCP clients to the local SSE URL. OPENAI_API_KEY is optional for OpenAI embeddings. CLI and Docker HTTP alternatives are documented.

Grounded Docs Cursor configuration

  1. Choose Cursor configuration scope

    For Grounded Docs MCP Server, open Project: .cursor/mcp.json User: ~/.cursor/mcp.json. Merge into mcpServers at project or user scope.

  2. Add Grounded Docs MCP Server to Cursor

    Use the docs-mcp-server entry below in mcpServers. Keep Grounded Docs MCP Server's documented sse connection fields and your other Cursor entries.

  3. Check Grounded Docs MCP Server in Cursor

    Grounded Docs MCP Server: Enable in Cursor’s MCP settings; inspect tools.

Config file · Project: .cursor/mcp.json

User: ~/.cursor/mcp.json

JSON
{
  "mcpServers": {
    "docs-mcp-server": {
      "url": "http://localhost:6280/sse"
    }
  }
}

Grounded Docs tools available to Cursor

Grounded Docs's stored tool list documents 1 tool; compare it with Cursor's connected tool list, because the stored list is not a live check.

scrape_docs

Find scrape_docs under Grounded Docs in Cursor's MCP settings; inspect its arguments when the agent requests a call.

Index documentation, with an optional setting to preserve hash-based routes.

Grounded Docs: full stored tool descriptions

Troubleshooting Grounded Docs in Cursor

  • Enable Grounded Docs MCP Server in Cursor's MCP settings and inspect its tools. For localhost:6280, check reachability, transport, and credential expiry if authentication fails.
  • Grounded Docs MCP Server: check sse connectivity to localhost:6280 from Cursor; a stdio command is not a replacement for this endpoint.
  • Transport mismatch: Grounded Docs's catalog lists stdio; the displayed Cursor configuration uses sse.
  • Known limitation: Hash-routed sites require explicit preservation settings and browser rendering.
  • Known limitation: Semantic vector search requires configuring an embedding model.
  • General Cursor troubleshooting lives in the /clients/cursor guide.

Grounded Docs MCP Server — details. Cursor guide

Frequently asked questions

Cursor's Grounded Docs MCP Server entry belongs under mcpServers in .cursor/mcp.json (project) or ~/.cursor/mcp.json (user). Confirm the active scope in MCP settings.

Cursor's docs-mcp-server example uses sse. The Grounded Docs MCP Server catalog lists stdio; these instructions follow the stored configuration instead.

Cursor connects Grounded Docs MCP Server to localhost:6280 using url.

Grounded Docs MCP Server's stored list contains 1 tool, including scrape_docs. Compare those names with Cursor's connected tool list; the stored list is not a live check.

Enable hash preservation explicitly. For fetch-mode jobs this can upgrade processing to Playwright so client-side routes can be evaluated.

Yes. The README shows indexing, searching, and single-page Markdown retrieval commands, with configurable structured output.