PluginBench
MCP Server
Active
MIT

io.github.docfork/docfork MCP Server

io.github.docfork/docfork

Up-to-date library docs for AI coding agents via MCP (service discontinued as of June 2026).

What is the io.github.docfork/docfork MCP server?

Docfork is an MCP server that provides AI coding agents (Cursor, Claude Code, Windsurf, and others) with up-to-date, searchable documentation for third-party libraries and frameworks. It exposes search_docs and fetch_doc tools so agents can retrieve current API references instead of relying on stale training data. Note: as of June 14, 2026, Docfork has shut down and the MCP server, CLI, and hosted endpoints are no longer functional.

Docfork was an MCP server designed to prevent AI coding agents from hallucinating APIs or using outdated library signatures by serving current documentation directly into tools like Cursor, Claude Code, and Windsurf. It offered two core tools — search_docs for ranked documentation results and fetch_doc for full markdown content — plus a CLI (dgrep) for terminal-based search and repo-scoped dependency resolution. Teams could index custom public or private GitHub repositories as searchable libraries and share access via API keys and "Cabinets." As of June 14, 2026, Docfork has shut down; the MCP server, CLI, and hosted endpoints are permanently offline and API keys no longer function.

How to install io.github.docfork/docfork

Copy-paste configuration for popular MCP clients.

transport: stdio
Config generated by PluginBench — verify against the source before use.
Claude Desktop
~/Library/Application Support/Claude/claude_desktop_config.json
{
  "mcpServers": {
    "docfork": {
      "command": "npx",
      "args": [
        "-y",
        "docfork"
      ]
    }
  }
}
Cursor
~/.cursor/mcp.json
{
  "mcpServers": {
    "docfork": {
      "command": "npx",
      "args": [
        "-y",
        "docfork"
      ]
    }
  }
}
Windsurf
~/.codeium/windsurf/mcp_config.json
{
  "mcpServers": {
    "docfork": {
      "command": "npx",
      "args": [
        "-y",
        "docfork"
      ]
    }
  }
}
VS Code
.vscode/mcp.json
{
  "servers": {
    "docfork": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "docfork"
      ]
    }
  }
}
Claude Code
claude mcp add docfork -- npx -y docfork

Tools & capabilities

Tools this server exposes to the agent.

  • search_docsReturns ranked documentation sections with titles, URLs, and relevance scores for a given library/query.
  • fetch_docReturns full rendered markdown content from a documentation URL found via search_docs.

Use cases

  • Retrieving current API signatures and usage examples for third-party libraries while coding
  • Setting up framework features (e.g., Next.js App Router SSR) using up-to-date docs instead of stale training data
  • Searching documentation from the terminal via the dgrep CLI
  • Indexing internal/private GitHub repositories (SDKs, runbooks) as custom searchable documentation libraries
  • Scoping documentation search to a project's declared dependencies via dgrep init

io.github.docfork/docfork MCP server FAQ

Is the Docfork MCP server still available?

No. Docfork shut down on June 14, 2026. The MCP server, CLI (dgrep), and all hosted endpoints are permanently offline, and existing API keys no longer work.

What did the Docfork MCP server do?

It served up-to-date documentation for libraries and frameworks directly into AI coding agents like Cursor, Claude Code, and Windsurf, via two tools: search_docs and fetch_doc.

Was Docfork free to use?

It offered a free tier of 1,000 requests per month per organization, with paid team plans for shared API keys and higher usage, before the service was shut down.

Did Docfork require an API key to use?

For individual use via `npx dgrep setup`, sign-in was required but no API key was needed; team/remote HTTP configurations used a DOCFORK_API_KEY header or OAuth.

How was Docfork different from Context7?

Docfork emphasized stack scoping (reading package.json to scope searches to declared dependencies), resolving package names once and caching them, and hybrid semantic + BM25 search with AST-aware chunking.

Did Docfork store user code or prompts?

According to the README, code and prompts never left the user's machine; only the query and library name were sent to Docfork, and private library content was end-to-end encrypted.

README (reference)

Source of truth, from the repository.

<p align="center"> <a href="https://docfork.com"> <picture> <source srcset="logo_light.png" media="(prefers-color-scheme: dark)"> <source srcset="logo_dark.png" media="(prefers-color-scheme: light)"> <img src="logo_dark.png" alt="Docfork" height="40" /> </picture> </a> </p>

[!CAUTION] Docfork has shut down as of June 14, 2026. The MCP server, CLI, and all hosted endpoints are offline and will not return. API keys no longer work, and npx dgrep setup will fail. Thank you to everyone who used the service.

<p align="center">Up-to-date docs for AI coding agents.</p>

AI agents hallucinate APIs, bloat context with stale docs, and write code against outdated signatures. Docfork serves up-to-date documentation directly in Cursor, Claude Code, and Windsurf.

Without Docfork

  app.use('/api/*', jwt({ secret: ... }))
-                       ^^^ removed in Hono v4

With Docfork

  app.use('/api/*', bearerAuth({ verifyToken: ... }))
+                       ^^^ current API, Hono v4.2

Get Started

npx dgrep setup

Installs the Docfork MCP server in your IDE. Detects your installed agents and writes the config file; sign in to Docfork on first use, no API key needed. Target one with --agent claude-code (also: cursor, codex, opencode, vscode, windsurf, amp, factory, zed).

Your agent now has two tools:

ToolReturns
search_docsRanked documentation sections with titles, URLs, and relevance scores.
fetch_docFull rendered markdown content from a documentation URL.

No prompt suffix needed:

Set up server-side rendering with Next.js App Router.

Or search from the terminal:

dgrep search "middleware redirect based on authentication" -l vercel/next.js
dgrep search "server actions with forms" -l vercel/next.js

Quickstart → · dgrep docs → · CLI reference →

Your own docs

Index any public or private GitHub repository as a custom library. Your internal APIs, SDKs, and runbooks become searchable by your agents — same pipeline as public libraries. GitHub integration setup →

Teams

Free: 1,000 requests/month per organization. For team rollout, commit the MCP config to your repo:

// .cursor/mcp.json (committed to git, picked up by every engineer)
{
  "mcpServers": {
    "docfork": {
      "url": "https://mcp.docfork.com/mcp",
      "headers": {
        "DOCFORK_API_KEY": "YOUR_TEAM_API_KEY"
      }
    }
  }
}

Share API keys and Cabinets across your organization. Docfork doesn't store your code or prompts. Security → · Pricing →

MCP Setup

[!TIP] Run npx dgrep setup to install automatically (use --agent claude-code to target one). Manual config below for other clients.

Cursor<a href="https://cursor.com/en/install-mcp?name=docfork&config=eyJ1cmwiOiJodHRwczovL21jcC5kb2Nmb3JrLmNvbS9tY3AifQ%3D%3D"><img src="https://cursor.com/deeplink/mcp-install-dark.svg" height="20" alt="Add to Cursor"/></a>

{
  "mcpServers": {
    "docfork": {
      "url": "https://mcp.docfork.com/mcp",
      "headers": {
        "DOCFORK_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Claude Code

claude mcp add --transport http docfork https://mcp.docfork.com/mcp/oauth

OpenCode

{
  "mcp": {
    "docfork": {
      "type": "remote",
      "url": "https://mcp.docfork.com/mcp",
      "headers": { "DOCFORK_API_KEY": "YOUR_API_KEY" },
      "enabled": true,
    },
  },
}

Don't see your client? Setup guides for all 29 supported clients →

OAuth Authentication

Docfork supports MCP OAuth specs. Change your endpoint to use OAuth:

- "url": "https://mcp.docfork.com/mcp"
+ "url": "https://mcp.docfork.com/mcp/oauth"

Note: OAuth is for remote HTTP connections only. View full OAuth guide →

Agent Rule

Add a rule so your agent calls Docfork MCP automatically. Full rule and IDE-specific setup →

[!NOTE] Add Rule to Cursor (One-Click)

Claude Code — add to your CLAUDE.md:

## Docfork policy

Use Docfork MCP `search_docs` and `fetch_doc` tools for library/API docs, setup, and configuration questions.

- Start `library` with a short name or keyword (e.g., `nextjs`, `zod`). Use the `owner/repo` from the result URL for follow-up calls, never guess it upfront.
- After finding a relevant result, call `fetch_doc` to get the full content. Search results are summaries only.
- Prefer Docfork results over training data when they conflict.
<details> <summary>Full rule (all clients)</summary>
When writing or debugging code that involves third-party libraries, frameworks, or APIs, use Docfork MCP `search_docs` and `fetch_doc` tools rather than relying on training data.

**Two defaults to follow every time:**

- Start `library` with a short name or keyword (e.g., `nextjs`, `zod`). Use the `owner/repo` from the result URL for follow-up calls, never guess it upfront.
- After finding a relevant result, call `fetch_doc` to get the full content. Search results are summaries only.

Skip Docfork when:

- Language built-ins, general algorithms, syntax stable across versions
- Code or docs the user has already provided in context

When uncertain, default to using Docfork.
</details>

FAQ

How is Docfork different from Context7? Both provide MCP servers and CLIs for searching library documentation. Here are the key differences:

  • Stack scoping. dgrep init reads your package.json and scopes all searches to your declared dependencies. Cabinets let you version-pin those libraries across a team.
  • Resolve once, search many. dgrep init resolves package names to canonical identifiers once and caches the mapping in .dgrep/config.json. No per-query resolution step.
  • Hybrid search. Semantic search and BM25 run in parallel, fused via Reciprocal Rank Fusion. AST-aware chunking preserves function boundaries.

Does Docfork store my code or prompts? Your code and prompts never leave your machine. At search time, only the query and library name are sent to Docfork — queries are not stored. Indexed documentation content lives in an upstream vector store; private library content is end-to-end encrypted and deleted atomically when you remove the library. Security →

What libraries are supported? Docfork maintains a curated catalog of popular frameworks. Add any public or private GitHub repository as a custom library. Add custom libraries →

Docs

Community

Star History

Star History Chart

License

MIT

Related MCP servers

Give your AI agent stealth web scraping with Cloudflare bypass and CSS selection, powered by Scrapling.

67k
Python
BSD-3-Clause
View repository →

Give your AI coding agent full control of a live Chrome browser for automation, debugging, and performance analysis.

45k
TypeScript
Apache-2.0
View repository →

Let AI agents manage your Puter files, websites, and serverless workers over MCP.

43k
TypeScript
AGPL-3.0
View repository →

Browser automation for AI agents via MCP, powering ByteDance's Agent TARS hybrid GUI/DOM browser control.

37k
TypeScript
Apache-2.0
View repository →

Run arbitrary shell commands from an MCP-connected AI agent.

37k
TypeScript
Apache-2.0
View repository →

Filesystem access MCP server from ByteDance's UI-TARS/Agent TARS ecosystem.

37k
TypeScript
Apache-2.0
View repository →