Local Research MCP Server

A free, private research assistant for AI development environments that searches the web and scrapes content using DuckDuckGo and Python. This MCP server gives AI assistants the ability to access real-time information while keeping all processing local for complete privacy.

- Claude Desktop- Anthropic's official desktop app
- Claude Code- Anthropic's CLI tool for developers
- Gemini CLI- Gemini's CLI tool
- Cline- VS Code extension for AI-powered coding
- Any MCP-compatible client

- 🔍Free Web Search: Uses DuckDuckGo's search API without requiring API keys
- 📄Content Extraction: Intelligently extracts clean text from web pages
- 🔒Privacy-First: All processing happens locally on your machine
- ⚡Fast Integration: Works seamlessly with Claude Desktop via MCP
- 🛠️Zero Dependencies: No external services or subscriptions required

- Python 3.10 or higher
- One of the compatible AI environments:

- Claude Desktop, Claude Code, Google CLI, Cline, or other MCP client

git clone https://github.com/Unlock-MCP/local-research-server.git cd local-research-server

uv init uv venv source .venv/bin/activate # On Windows: .venv\Scripts\activate

uv add "mcp[cli]" duckduckgo-search trafilatura

- macOS:~/Library/Application Support/Claude/claude_desktop_config.json
- Windows:%APPDATA%/Claude/claude_desktop_config.json

{ "mcpServers": { "local-researcher": { "command": "uv", "args": [ "--directory", "/path/to/your/local-research-server", "run", "python", "research_server.py" ] } } }

Add this to yourCLAUDE.mdfile or use the CLI:

claude --mcp-server local-researcher="uv --directory /path/to/your/local-research-server run python research_server.py"

Configure via your MCP client settings or environment variables.

Add to your Cline MCP server configuration in VS Code settings.

Replace/path/to/your/local-research-serverwith the full path to this directory.

After configuration, restart your AI environment. Connection indicators vary by platform:

- Claude Desktop: Look for the plug icon (🔌)
- Claude Code: Server status shown in CLI output
- Other platforms: Check your MCP client's connection status

Ask your AI assistant research questions like:

- "What are the latest developments in AI research this week?"
- "Research current renewable energy trends in Europe"
- "Find recent cybersecurity threat reports"

Your AI assistant will automatically use the local server to fetch real-time information.

The server implements a single MCP tool that:
- Searchesthe web using DuckDuckGo's free API
- Scrapescontent from the top search results
- Extractsclean, readable text using Trafilatura
- Returnsformatted content to Claude for analysis

All processing happens locally with no data sent to external services except for the public web searches.

AI Client (Claude Desktop/Code/Gemini/Cline) ↓ (MCP Protocol) Local Research Server ↓ (Search API) DuckDuckGo Search ↓ (HTTP Requests) Target Websites ↓ (Content Extraction) Clean Text → AI Assistant

The server accepts these parameters for the research tool:

- query(string): The search query or research topic
- num_results(int): Number of search results to process (default: 3)

- Input Validation: Sanitizes search queries
- Rate Limiting: Polite delays between web requests
- Error Handling: Graceful failure handling
- Local Processing: No external data dependencies

- Content Creation: Research-backed article writing
- Market Analysis: Real-time competitor and industry research
- Due Diligence: Company and investment research
- Regulatory Monitoring: Tracking compliance requirements

- Add RSS feed integration
- Implement search result caching
- Add domain filtering capabilities
- Include publication date extraction
- Add multiple search engine support

- Verify the absolute path in your AI client config
- Restart your AI environment completely
- Check that Python dependencies are installed

- Verify internet connection
- Check DuckDuckGo service status
- Try different search queries

- Some websites block scraping
- Try different search terms for varied sources
- Check the console output for specific errors

We welcome contributions! Please see ourContributing Guidefor details.

- MCP Docs Server- Local documentation access
- Remote MCP Server- Enterprise database access
- UnlockMCP Website- MCP tutorials and resources

- Tutorial Guide
- MCP Documentation
- Issue Tracker

Search the web using Kagi's search API