Documentation Search

You have an MCP server that lets you search documentation across many sources, scan Python projects for vulnerabilities, and generate project boilerplates. It’s accessible from MCP clients and can be extended with optional live web search and semantic capabilities, making it a powerful helper for development workflows.

How to use

You interact with the Documentation Search MCP Server through an MCP client. Start by ensuring the server is running via the specified command, then use the provided MCP tools to search documentation, fetch docs, generate learning paths, or create project boilerplates. You can optionally enable live web search with an API key and turn on semantic search for richer results. When you issue a request, you’ll typically specify the tool (for example, semantic_search or get_docs) and provide any required parameters. The MCP server returns results you can inspect in your client’s UI or workflow.

How to install

Prerequisites: you need Python installed and you should have a modern MCP client setup. You also have the option to install via uvx for a smoother experience.

# Recommended: install via uvx (install uv from an official source if needed)
uvx documentation-search-enhanced@1.9.0

# Or install in a Python virtual environment
pip install documentation-search-enhanced==1.9.0

# Optional: enable AI semantic search (Python 3.12 only; adds ~600MB)
pip install documentation-search-enhanced[vector]==1.9.0

Configuration

Configure your MCP client to connect to the server using a stdio-based setup, or connect via an MCP management CLI. The following example demonstrates how to register the server with a client-side bridge and optionally enable a live web search API key.

{
  "mcpServers": {
    "documentation_search_enhanced": {
      "command": "/Users/yourusername/.local/bin/uvx",
      "args": ["documentation-search-enhanced@1.9.0"],
      "env": {
        "SERPER_API_KEY": "optional_key_here"
      }
    }
  }
}

Usage notes

If you want to run with live web search, you can supply a SERPER_API_KEY. Without it, the server uses a prebuilt index shipped with releases. You can disable live crawling by setting the appropriate feature flag in your config. If you see timeouts on first run, pre-download dependencies by starting the server once to fetch assets.

Additional configuration and options

Environment variables you may encounter and how they affect behavior include SERPER_API_KEY for live web search, DOCS_SITE_INDEX_AUTO_DOWNLOAD to control automatic index downloads, and DOCS_SITE_INDEX_PATH to set a custom path for the documentation index.

Available tools

semantic_search

Search documentation with optional vector-based semantic ranking and keyword matching.

get_docs

Fetch specific documentation pages or sources.

get_learning_path

Generate a learning roadmap tailored to your goals.

get_code_examples

Find code snippets relevant to your search queries.

scan_project_dependencies

Scan a Python project’s dependencies for known vulnerabilities.

snyk_scan_project

Provide detailed Snyk analysis for a project.

generate_project_starter

Create starter boilerplates for FastAPI and React projects.

manage_dev_environment

Generate docker-compose files and development environments.

compare_library_security

Compare the security of libraries across options.