Changelog

All notable changes to ai-pr-describer will be documented in this file

v1.0.0

Go Rewrite: Efficient Go implementation.
Universal AI Model Support: Introduced openai-base-url for compatibility with any OpenAI-compatible API (ChatGPT, DeepSeek, Groq, local Ollama, etc.).
PR Description Auto-Update: The action now directly updates the Pull Request description (body) with the AI-generated summary.
Modern Infrastructure: Switched to a multi-stage Dockerfile for smaller images and Go modules for dependency management.
Improved Prompting: Redesigned the AI prompt for more structured, information-dense summaries.

High-Performance Execution: Transitioned from a Docker-based action to a Composite Action using pre-built binaries. Initial startup time is significantly reduced from ~2 minutes to under 15 seconds.
Configurable Limits:
- max-tokens: Control the length of the generated PR summary (default: 2000).
- max-context-tokens: Adjust the total context window size (default: 256,000, optimized for large models like GLM).
Automated Releases: Added a GitHub Actions workflow to automatically build and release binaries for various platforms (Linux, macOS) on tag push.
Enhanced Debugging: Integrated ::debug:: logs to help troubleshoot and monitor execution times for different stages.

AI Summary Quality: Updated the core prompt format to include specialized sections like 🚀 Summary, 🔑 Key Changes, and ⚠️ Breaking Changes.
Documentation: Comprehensive update to README.md with modern examples and performance tips.