Context42

MCP server for semantic search over your personal coding instructions.

What it does

Context42 indexes your markdown documentation and makes it searchable by AI assistants like Claude. Your coding standards, guidelines, and patterns become instantly accessible during development.

Installation

# Using pipx (recommended)
pipx install context42-io

# Using uv
uvx context42-io

# Using pip
pip install context42-io

Quick Start

# 1. Add your documentation
c42 add ~/my-coding-standards --name standards

# 2. Index the content
c42 index

# 3. Start the MCP server
c42 serve

Claude Desktop Integration

Add to your Claude Desktop configuration file:

{
  "mcpServers": {
    "context42": {
      "command": "c42",
      "args": ["serve"]
    }
  }
}

Restart Claude Desktop after adding the configuration.

CLI Commands

Priority System

Set higher priority for your personal instructions so they take precedence over reference documentation:

# Your rules (high priority)
c42 add ~/my-standards --name standards --priority 1.0

# Reference docs (lower priority)
c42 add ~/library-docs --name docs --priority 0.5

Search results are weighted by priority, ensuring your preferences appear first.

Features

• 100% Offline - Runs locally after initial model download
• Priority System - Your rules take precedence over reference docs
• Fast Indexing - Progress bars with batch processing
• Idempotent - Safe to re-run indexing

Configuration

Environment variables (optional):

C42_EMBEDDING_MODEL="BAAI/bge-small-en-v1.5"  # Default model
C42_CHUNK_SIZE=500                             # Characters per chunk
C42_BATCH_SIZE=50                              # Chunks per batch
C42_DATA_DIR="~/.local/share/context42"        # Data directory

Advanced Configuration

For users who need faster model downloads or encounter rate limiting:

# Optional: Set HuggingFace token for faster downloads
export HF_TOKEN="hf_your_token_here"

Get your token at: https://huggingface.co/settings/tokens

Data Storage

Context42 stores data in platform-specific locations:

License

MIT