llm-codepac 0.1.0

Generate structured Context Packs for AI coding tools by analyzing your codebase.

ContextCraft

A Python CLI tool that analyzes a code repository and generates a structured Context Pack document optimized for injecting into AI coding tools (Claude, GPT, Cursor, Copilot) so they understand the codebase before writing code.

Installation

CLI

pip install contextcraft

VS Code Extension

Coming soon on the VS Code Marketplace.

Usage

export ANTHROPIC_API_KEY=your_key_here
contextcraft init /path/to/your/repo

Environment

Create a .env file (see .env.example) or set:

• ANTHROPIC_API_KEY — required for AI synthesis (omit or use --no-ai for offline)

Output

• Default: context.pack.md in the repo root with a Quick Copy section and full Context Pack.
• JSON: Raw analysis or pack as JSON when --format json is used.

Requirements

• Python 3.11+
• Git (optional; used for recent commits and hotspot analysis)

Distribution

• Install from PyPI: pip install contextcraft
• Install the VS Code extension: ContextCraft on the Marketplace (when published), or install manually: download the .vsix from Releases and run Extensions: Install from VSIX... in VS Code.
• Publish a new Python release: tag with git tag v0.x.x && git push --tags (triggers PyPI publish).
• Publish a new extension release: tag with git tag ext-v0.x.x && git push --tags (triggers VS Code Marketplace publish and uploads .vsix to the GitHub Release).