Docs MCP Server

Active
GitHub TypeScript MIT

Description

Grounded Docs MCP Server providing precise technical documentation retrieval for AI coding assistants, an open-source alternative to Context7.

Key Features

  • Fetches documentation from websites, GitHub repos, npm, PyPI, and local files on demand
  • Version-specific queries targeting exact library versions in your project
  • Supports 90+ source code languages plus HTML, Markdown, PDF, Office docs, and archives
  • Optional semantic embedding models (OpenAI, Ollama, Gemini) for vector-based search
  • Runs entirely local — your code and data never leave your machine
  • Web UI for managing documentation indexes, compatible with all MCP clients

Use Cases

💡 Providing AI coding assistants with up-to-date, version-specific library documentation
💡 Reducing LLM hallucinations by grounding responses in real official documentation
💡 Indexing and searching across multiple documentation sources in a unified interface
💡 Integrating with Claude, Cline, Copilot, or any MCP-compatible coding assistant
💡 Scraping and caching documentation from SPA sites with hash-based routing

Tags

typescript mcp tools coding documentation

Categories

🔌 MCP Tools

Quick Start

npx @arabold/docs-mcp-server@latest scrape <name> <url> → npx @arabold/docs-mcp-server@latest search <name> '<query>' → Optional: start MCP server with npx → Open Web UI at localhost:6280 → Connect MCP client via SSE endpoint → Enable embedding model for semantic search
Visit GitHub

Related Projects