meshdb-go 2.0.0

Local full-text + vector search with retrieval-first philosophy

MeshDB

Local full-text + vector search with retrieval-first philosophy.

Installation

pip install meshdb-go

Quick Start

# Check status
meshdb status

# Index your files
meshdb index ~/Documents

# Search
meshdb search "meeting notes"

# Semantic search (requires Ollama)
meshdb semantic "how does authentication work"

Features

• Full-text search (FTS5) with multilingual support
• Semantic search using Ollama embeddings (bge-m3)
• Hybrid search combining keyword + vector similarity
• MCP integration for Claude Desktop
• 100% local - no data leaves your machine
• Sensitive data blocking at indexing time

Requirements

• macOS or Linux
• Python 3.9+
• Ollama (optional, for semantic search)

Ollama Setup (Optional)

For semantic search capabilities:

# Install Ollama
brew install ollama  # macOS
# or: curl -fsSL https://ollama.com/install.sh | sh  # Linux

# Start Ollama
ollama serve

# Pull embedding model
ollama pull bge-m3

Claude Desktop Integration

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

{
  "mcpServers": {
    "meshdb": {
      "command": "meshdb",
      "args": ["mcp"]
    }
  }
}

Documentation

• Quick Start
• Installation Guide
• MCP Integration

License

MIT