AutoRebase — Automated Re-targeting of Feature Customizations

Overview

AutoRebase helps re-target feature customizations when a Base SW updates. Given base-OLD, base-NEW, and feature-OLD, it produces feature-NEW with re-applied customizations, machine-readable patchsets, and a human report with traceability and validation.

🚀 Quick Start Options

Option 1: MCP Server (Recommended for AI Integration)

The MCP (Model Context Protocol) server provides AI-powered conflict resolution and is perfect for integration with AI models like Cursor.

Setup:

cd mcp-server
npm install
npm run build

Usage with Cursor:

1. Copy mcp-server/cursor-mcp-config.json to your Cursor MCP configuration
2. Restart Cursor
3. Use AI to call AutoRebase tools directly

Available Tools:

• autorebase: Complete AutoRebase process with AI conflict resolution
• validate_repository: Validate GitHub repositories and SHAs

See mcp-server/README.md for detailed MCP setup instructions.

Option 2: FastAPI Server

REST API for programmatic access to AutoRebase functionality.

Setup:

pip install -r api/requirements.txt
python api/main.py

Usage:

curl -X POST "http://localhost:8000/autorebase" \
  -H "Content-Type: application/json" \
  -d '{
    "base_software_0": "main",
    "base_software_1": "main", 
    "feature_software_0": "main",
    "base_repo_url": "https://github.com/user/base-repo.git",
    "feature_repo_url": "https://github.com/user/feature-repo.git"
  }'

See README_FASTAPI.md for detailed API documentation.

Option 3: Engine CLI (Legacy)

Command-line interface for direct AutoRebase operations.

Setup:

bash scripts/install_tools.sh

Usage:

python -m engine.cli.auto_rebase init --old-base data/sample/base-1.0 --new-base data/sample/base-1.1 --feature data/sample/feature-5.0 --req-map data/sample/requirements_map.yaml --workdir artifacts/run1
python -m engine.cli.auto_rebase extract-feature --out artifacts/run1/feature_patch
python -m engine.cli.auto_rebase extract-base --out artifacts/run1/base_patch
python -m engine.cli.auto_rebase retarget --feature-patch artifacts/run1/feature_patch --base-patch artifacts/run1/base_patch --new-base data/sample/base-1.1 --out artifacts/run1/feature-5.1
python -m engine.cli.auto_rebase validate --path artifacts/run1/feature-5.1 --report artifacts/run1/report.html
python -m engine.cli.auto_rebase finalize --path artifacts/run1/feature-5.1 --tag v5.1 --trace artifacts/run1/trace.json

🎯 What You Get

• feature-NEW tree with applied changes
• Machine-readable ΔF and ΔB patchsets
• Complete changelog with detailed processing information
• AI-powered conflict resolution using OpenAI
• Requirements-based resolution respecting REQUIREMENTS_MAP.yml
• report.html and report.json (schema-validated)
• trace.json listing requirement IDs per patch unit
• Optional git tag and commit trailers

🏗️ Architecture

MCP Server (mcp-server/)

• Model Context Protocol server for AI integration
• TypeScript/Node.js implementation
• Provides tools for AI models to call AutoRebase functionality
• Supports local SSH authentication and GitHub token authentication

FastAPI Server (api/)

• REST API for programmatic access
• Python implementation with Pydantic models
• Async/await support for concurrent operations
• Comprehensive error handling and validation

Engine (engine/)

• Core AutoRebase logic and algorithms
• Adapter pattern for different file types
• Fallback strategies for missing external tools
• Python implementation with extensive testing

Web Interface (web/)

• Next.js web application
• User-friendly interface for AutoRebase operations
• Real-time progress tracking and results visualization

🔧 Environment Requirements

• Python: 3.11+ recommended
• Node.js: 18+ for MCP server and Next.js app
• OpenAI API Key: Required for AI conflict resolution
• Git: Required for repository operations
• Optional external tools: difftastic, gumtree, clang-tidy, coccinelle, dtc, yq, comby

🚀 Key Features

AI-Powered Conflict Resolution

• Uses OpenAI GPT models to intelligently resolve merge conflicts
• Respects REQUIREMENTS_MAP.yml for context-aware resolution
• Validates resolutions with confidence scoring
• Handles complex multi-file conflicts

Automatic Changelog Generation

• Complete processing history with timestamps
• Detailed file-by-file change tracking
• Patch application success/failure logging
• AI resolution statistics and validation scores

Flexible Authentication

• SSH: Uses local SSH keys for GitHub operations
• GitHub Token: Supports personal access tokens
• Public Repositories: No authentication required for read operations
• Local Credentials: Falls back to system Git configuration

Robust Error Handling

• Graceful degradation when external tools are missing
• Comprehensive validation of inputs and outputs
• Detailed error messages with troubleshooting guidance
• Fallback strategies for edge cases

🧪 Development

• Formatting/lint: ruff, black, isort configured in pyproject.toml
• Tests: pytest (see tests/)
• CI: GitHub Actions in .github/workflows/ci.yml
• Type Safety: TypeScript for MCP server, Pydantic for Python APIs

🔒 Security & Privacy

• Local Processing: All operations happen on your machine
• No Data Storage: No sensitive data is stored or transmitted
• Secure Authentication: Multiple authentication methods supported
• API Key Management: Secure handling of OpenAI and GitHub tokens
• Public Repository Focus: Designed for open-source workflows

📚 Documentation

• MCP Server: mcp-server/README.md - AI integration guide
• FastAPI: README_FASTAPI.md - REST API documentation
• Cursor Setup: mcp-server/CURSOR_SETUP.md - Cursor AI integration
• Engine: engine/README.md - Core algorithm documentation

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test thoroughly (unit tests + integration tests)
5. Submit a pull request

📄 License

MIT License - see LICENSE file for details.