Show HN: Git Auto Commit (GAC) – LLM-powered Git commit command line tool (github.com)

Git Auto Commit (GAC) is a new CLI that replaces manual git commit messages with LLM-generated, context-aware messages. After a simple gac init to configure an LLM provider (supports OpenAI, Anthropic, Gemini, Ollama, LM Studio, Groq, Cerebras, and many others), running gac analyzes your staged diffs, infers intent (refactor, bugfix, feature, breaking change), and produces well-structured messages you can accept, reroll, or tweak. It offers one-line conventional commits (-o), standard summaries, and verbose output (-v) with Motivation, Technical Approach, and Impact Analysis, plus interactive hints (-h), scope inclusion (-s), and a reroll command for iterative refinement. Technically it integrates with git hooks (respects pre-commit and lefthook), smart-filters generated files/dependencies, and includes automatic secret detection and interactive prompts for sensitive data; you can skip the scan with --skip-secret-scan. Workflows support staging, auto-confirm, and push in one command (gac -ayp), and debugging features like --show-prompt reveal what the model sees. For teams this can standardize commit quality and save time, but it raises privacy considerations when sending diffs to remote models—GAC mitigates this with local model support and explicit prompts about secrets. Documentation and env-variable config (.gac.env.example, USAGE.md) are provided for quick setup.