repo2ai 0.2.3

Export Git repository contents to structured Markdown with AI chat integration

Repo2AI

Export Git repositories to structured Markdown with AI Chat Integration. Perfect for code analysis, documentation, and getting AI assistance with your projects.

Quick Start

pip install repo2ai

# Basic export to stdout
repo2ai .

# Copy to clipboard and open ChatGPT
repo2ai . --open-chat chatgpt

# Open Claude with a specific prompt
repo2ai . --open-chat claude --prompt "Analyze this codebase and suggest improvements"

Features

• Smart Repository Scanning - Respects .gitignore, filters binary files, handles large codebases
• AI Chat Integration - Automatic browser integration with ChatGPT, Claude, and Gemini
• Scope Filtering - Focus on recent commits, uncommitted changes, or specific file patterns
• Multiple Output Options - File, clipboard, stdout, or combinations
• Intelligent Filtering - Configurable exclude patterns, file size limits, meta file handling
• Language Detection - Syntax highlighting for 25+ programming languages
• Git Integration - Only includes tracked files, respects Git ignore patterns

Usage Examples

Basic Export

repo2ai .                              # Export to stdout
repo2ai ./project --output docs.md     # Export to file
repo2ai . --clipboard                  # Copy to clipboard
repo2ai . -s                           # Short summary output

AI Chat Integration

Open your repository directly in ChatGPT, Claude, or Gemini:

repo2ai . --open-chat chatgpt
repo2ai . --open-chat claude --prompt "Review for security vulnerabilities"
repo2ai . --open-chat gemini --prompt "Help me write documentation"
repo2ai . --chat-all --prompt "Explain this codebase"
repo2ai . --open-chat chatgpt --browser firefox

Supported services: chatgpt, claude, gemini

PR Review Mode

Generate AI-friendly context for code reviews. This mode is specifically optimized to give an AI everything it needs to understand your changes: the metadata, the actual diff, and the full content of every changed file.

1. Automatic Target Detection

If you don't specify a target branch, repo2ai automatically detects the best merge target:

• It checks for the remote default branch (origin/HEAD).
• It looks for common base branches (main, master, develop, dev).
• It uses the upstream tracking branch if it's different from your current branch.

# Auto-detect target (e.g., compares feature → main)
repo2ai . --pr-review

2. Specific Targets

You can compare against any local or remote branch:

# PR context against specific local branch
repo2ai . --pr-review develop

# PR context against remote main branch
repo2ai . --pr-review origin/main

# PR review with Claude and a custom prompt
repo2ai . --pr-review --open-chat claude --prompt "Analyze for logic errors and performance"

3. What's Included?

The generated Markdown report contains:

• Summary: Source and target branches, number of changed files, and total commit count.
• Unified Diff: The exact code changes for concise review.
• Full File Context: The complete current content of every changed file, enabling the AI to understand the impact of your changes on the surrounding code.

Scope Filtering (Focused Analysis)

Limit export to specific files for focused AI analysis:

# Files from last 3 commits
repo2ai . --recent 3

# Only uncommitted changes
repo2ai . --uncommitted

# All Python files
repo2ai . --include "**/*.py"

# Combined: recent Python changes in src/
repo2ai . --recent 1 --include "src/**/*.py"

# Python files, excluding tests
repo2ai . --include "**/*.py" --exclude "**/test_*.py"

Advanced Filtering

# Exclude sensitive files
repo2ai . --exclude "*.env" --exclude "secrets/*"

# Large repositories with size limits
repo2ai . --max-file-size 50000 --no-meta

# Custom exclude patterns
repo2ai . --exclude "*.log" --exclude "temp/*"

CLI Options

Run repo2ai --help for complete options. Key flags:

Option	Description
--output, -o	Save to file
--clipboard, -c	Copy to clipboard
--stdout, -s	Output to stdout (default)
--open-chat SERVICE	Open AI service (chatgpt, claude, gemini)
--chat-all	Open all AI services
--prompt TEXT	Initial prompt for AI service
--browser BROWSER	Browser to use (chrome, firefox, safari, edge)
--exclude PATTERN	Exclude files matching pattern
--no-meta	Exclude README, LICENSE, etc.
--max-file-size BYTES	Maximum file size (default: 1MB)
--recent N	Only files from last N commits
--uncommitted	Only uncommitted changes
--include PATTERN	Only files matching glob pattern
--pr-review [TARGET]	Generate PR review context (diff + changed files)
-v, --verbose	Show detailed file lists

Contributing

git clone https://github.com/georghildebrand/repo2ai.git
cd repo2ai

# Setup
make setup         # Install dependencies + dev tools + pre-commit hooks
make install       # Install dependencies only

# Development
make format        # Auto-format code with Black
make lint          # Run flake8 + mypy
make test          # Run tests
make test-cov      # Run tests with coverage report

# Validation
make ci            # Run all checks (format-check, lint, test)
make check         # Same as ci
make install-system # Install built wheel into user site-packages (dev-only)
make uninstall-system # Remove system installation

# Release & Versioning
make version       # Show current project version
make release PART=patch  # Level release (patch, minor, or major)

# Build & Clean
make build         # Build wheel and sdist
make clean         # Remove build artifacts and caches
make distclean     # Deep clean including virtualenv

# Full pipeline
make all           # Run check + docs + build

Releasing a New Version

The project uses automated GitHub Actions for releases. To trigger a new release:

1. Run make release PART=patch (or minor/major).
2. The command will automatically:
   • Run the quality gate (make check).
   • Bump the version in pyproject.toml.
   • Create a Git tag (e.g., v0.2.1).
   • Push the tag and code to GitHub.
3. GitHub Actions will then build the package, publish it to PyPI, and create a GitHub Release with documentation.

See Contributing Guidelines for details.

Architecture

The tool follows a modular architecture: CLI parsing, repository scanning, scope filtering, PR review context generation, markdown generation, and output handling. See Architecture Documentation for details.

License

MIT License - see LICENSE for details.

Links

• GitHub Repository
• PyPI Package
• Architecture Diagrams