Research Assistant MCP Server

A Model Context Protocol (MCP) server that integrates Wikipedia as a knowledge source with Obsidian for personal knowledge management. Run locally with Claude Desktop for seamless research and note-taking workflows.

Features

• 🔍 Wikipedia Integration: Search and fetch full articles from Wikipedia
• 📝 Obsidian Notes: Create and search notes with wikilinks and backlinks
• 🎯 Research Modes: Configure research rigor (strict/balanced/exploratory)
• 🔗 Smart Linking: Automatic wikilink generation and backlink tracking
• 💾 Local-First: Everything runs locally, no external dependencies beyond Wikipedia API

Quick Start

Prerequisites

• Python 3.10 or higher
• An Obsidian vault
• Claude Desktop

Installation

1. Clone or navigate to the project directory:

   cd research-mcp-server

2. Create and activate a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install the package:

   pip install -e .

4. Set up environment variables:

   cp .env.example .env
   # Edit .env and set your Obsidian vault path

Configuration

Environment Variables

Create a .env file or set these in your Claude Desktop configuration:

OBSIDIAN_VAULT_PATH=/path/to/your/ObsidianVault
CACHE_ENABLED=true
DEFAULT_RESEARCH_MODE=balanced

Claude Desktop Configuration

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS):

{
  "mcpServers": {
    "research-assistant": {
      "command": "python",
      "args": ["-m", "research_mcp_server"],
      "env": {
        "OBSIDIAN_VAULT_PATH": "/Users/yourusername/Documents/ObsidianVault"
      }
    }
  }
}

After editing:

1. Save the file
2. Restart Claude Desktop
3. The server will auto-start when Claude launches

Available Tools

1. search_wikipedia

Search Wikipedia for articles matching a query.

Parameters:

• query (string, required): Search query
• limit (integer, optional): Maximum results (default: 5)

Example:

Search Wikipedia for "consistent hashing"

2. fetch_wikipedia_article

Fetch full Wikipedia article content.

Parameters:

• title (string, required): Article title

Example:

Fetch the Wikipedia article on "Distributed Hash Table"

3. create_obsidian_note

Save a note to your Obsidian vault with wikilinks and backlinks.

Parameters:

• title (string, required): Note title
• content (string, required): Note content in markdown
• tags (array, optional): Tags for the note
• links (array, optional): Related notes to link

Example:

Create a note titled "Consistent Hashing" with content about the algorithm

4. search_obsidian_notes

Search notes in your Obsidian vault.

Parameters:

• query (string, required): Search query

Example:

Search my notes for "distributed systems"

5. set_research_mode

Configure research rigor and citation style.

Parameters:

• mode (string, required): Research mode - strict, balanced, or exploratory
• citation_style (string, optional): Citation style - inline, footnoted, or minimal

Research Modes:

• Strict: Only use information from fetched sources, cite everything
• Balanced: Prioritize sources but add context from general knowledge
• Exploratory: Use sources as starting points, make broader connections

Example:

Set research mode to strict with inline citations

Usage Examples

Research Workflow

1. Search for a topic:

   Search Wikipedia for "CAP theorem"
   

2. Fetch detailed article:

   Fetch the Wikipedia article on "CAP theorem"
   

3. Create a note:

   Create a note about CAP theorem with tags "distributed-systems" and "theory"
   

4. Link related concepts:

   Create a note linking to "Consistency Models" and "Distributed Consensus"
   

5. Search your notes:

   Search my notes for "consistency"
   

Project Structure

research-mcp-server/
├── pyproject.toml              # Project configuration
├── README.md                   # This file
├── .env.example                # Environment template
├── src/
│   └── research_mcp_server/
│       ├── __init__.py
│       ├── __main__.py         # Entry point
│       ├── server.py           # MCP server implementation
│       ├── config.py           # Configuration loader
│       ├── models.py           # Data models
│       ├── research_mode.py    # Research mode manager
│       ├── adapters/
│       │   ├── __init__.py
│       │   ├── base.py         # Abstract adapter
│       │   └── wikipedia.py    # Wikipedia implementation
│       └── obsidian/
│           ├── __init__.py
│           └── vault.py        # Obsidian operations
└── tests/
    ├── test_wikipedia.py
    ├── test_obsidian.py
    └── test_server.py

Development

Running Tests

pip install -e ".[dev]"
pytest

Code Formatting

black src/
ruff check src/

Known Limitations

• Platform Support: Tested on macOS and Linux only (not Windows)
• Concurrency: Handles requests serially (MCP SDK manages this)
• Cache: In-memory only, not persisted between sessions
• Language: English Wikipedia only
• Authentication: None (local-only operation)

Troubleshooting

Server doesn't start

1. Check that OBSIDIAN_VAULT_PATH is set correctly
2. Verify the vault path exists
3. Check Claude Desktop logs for errors

Wikipedia requests fail

• Check your internet connection
• Wikipedia API may be rate-limiting (server retries automatically)

Notes not appearing in Obsidian

• Ensure the vault path is correct
• Check file permissions
• Refresh Obsidian vault

Philosophy

This project follows these principles:

• Simplicity First: Use abstractions only where they enable future extensibility
• Working Software: Prioritize functional, reliable code over architectural perfection
• Pragmatic Modularity: Clean boundaries between components, no premature generalization
• Fail Loudly: Clear error messages, obvious failure modes
• Local-First: Everything runs locally, no external dependencies beyond Wikipedia API

Future Enhancements

Potential additions (not in MVP):

• Additional knowledge sources (arXiv, PubMed, etc.)
• Persistent caching
• Advanced search with filters
• Note templates
• Automatic backlink updates

License

MIT License - feel free to use and modify as needed.

Contributing

This is a personal project, but suggestions and improvements are welcome!