obsidian-notes-rag 0.2.0

MCP server for semantic search over Obsidian notes using local RAG

obsidian-rag

MCP server for semantic search over your Obsidian vault. Uses OpenAI embeddings by default (or Ollama for local processing) with ChromaDB for vector storage.

What it does

Ask natural language questions about your notes:

• "What did I write about project planning?"
• "Find notes similar to my meeting notes from last week"
• "What's in my daily notes about the API refactor?"

Requirements

• Python 3.11+
• OPENAI_API_KEY environment variable (or Ollama for local embeddings)

Quick Start

The easiest way to get started is with uvx (no installation required):

# Run the setup wizard
uvx obsidian-notes-rag setup

Add to Claude Code (CLI)

claude mcp add -s user obsidian-notes-rag -- uvx obsidian-notes-rag serve

Add to Claude Desktop (JSON config)

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "obsidian-notes-rag": {
      "command": "uvx",
      "args": ["obsidian-notes-rag", "serve"]
    }
  }
}

Alternative: Clone and install

git clone https://github.com/ernestkoe/obsidian-rag.git
cd obsidian-rag
uv sync

uv run obsidian-notes-rag setup
claude mcp add -s user obsidian-notes-rag -- uv run --directory /path/to/obsidian-rag obsidian-notes-rag serve

The setup wizard will:

1. Ask for your embedding provider (OpenAI or Ollama)
2. Configure your API key (for OpenAI)
3. Set your Obsidian vault path
4. Choose where to store the search index
5. Optionally run the initial indexing

Manual Setup (alternative)

# Set your API key and index directly
export OPENAI_API_KEY=sk-...
uv run obsidian-notes-rag index --vault /path/to/your/vault

Using Ollama (local, offline)

# Install Ollama and pull the embedding model
ollama pull nomic-embed-text

# Run setup with Ollama, or index directly:
uv run obsidian-notes-rag --provider ollama index --vault /path/to/your/vault

MCP Tools

Once connected, these tools are available to Claude:

Tool	What it does
search_notes	Find notes matching a query
get_similar	Find notes similar to a given note
get_note_context	Get a note with related context
get_stats	Show index statistics
reindex	Update the index

Keeping the Index Fresh

Option 1: Manual reindex

uv run obsidian-notes-rag index

Option 2: Watch for changes

uv run obsidian-notes-rag watch

Option 3: Auto-start on login (macOS)

uv run obsidian-notes-rag install-service

CLI Reference

obsidian-notes-rag setup                # Interactive setup wizard
obsidian-notes-rag serve                # Start MCP server (for Claude Code)
obsidian-notes-rag index [--clear]      # Index vault (--clear to rebuild)
obsidian-notes-rag search "query"       # Search from command line
obsidian-notes-rag watch                # Watch for file changes
obsidian-notes-rag stats                # Show index stats
obsidian-notes-rag install-service      # Install macOS launchd service
obsidian-notes-rag uninstall-service    # Remove service
obsidian-notes-rag service-status       # Check service status

Configuration

Set your vault path and provider via CLI options or environment variables:

# CLI options
uv run obsidian-notes-rag --vault /path/to/vault index
uv run obsidian-notes-rag --provider ollama index

# Environment variables
export OBSIDIAN_RAG_VAULT=/path/to/vault
export OBSIDIAN_RAG_PROVIDER=ollama  # or "openai" (default)

Variable	Description
OPENAI_API_KEY	OpenAI API key (required for default provider)
OBSIDIAN_RAG_PROVIDER	Embedding provider: openai (default) or ollama
OBSIDIAN_RAG_VAULT	Path to Obsidian vault
OBSIDIAN_RAG_DATA	Where to store the index (default: ./data)
OBSIDIAN_RAG_OLLAMA_URL	Ollama API URL (default: http://localhost:11434)
OBSIDIAN_RAG_MODEL	Override embedding model

How it works

1. Parses your markdown files and splits them by headings
2. Generates embeddings using OpenAI API (or Ollama for local processing)
3. Stores vectors in ChromaDB (local, persistent)
4. MCP server provides semantic search to Claude

Contributing

See CONTRIBUTING.md for development setup.

License

MIT