Released my first open source project this week and wanted to share it here in case it's useful or if anyone wants to contribute.
The project: aicontext-commit — a small CLI that generates commit messages by sending your staged diff and your last 30 commits to Claude or OpenAI. The history acts as a style reference so the suggestions match your repo's existing conventions (language, Conventional Commits, casing, level of detail).
Why open source: the alternatives I found were either closed source, locked into one IDE, or didn't expose the prompt. I wanted something I could fork, audit the prompt of, and run alongside any LLM provider. MIT made that easy for anyone.
Stack: TypeScript compiled to CommonJS, no bundler. Tests run via node:test against the compiled output. ~200 lines of source split across git, prompt, ai, and errors modules.
State of the project: published to npm, working end-to-end with both Anthropic and OpenAI APIs. Feature surface is intentionally small. Things I'd love help with:
Local model support (Ollama integration, probably)
Branch-name and PR-title as additional context
Tests for edge cases (very large diffs, weird commit history, binary files)
Documentation improvements (especially translations)
Maintenance commitment: I'll keep responding to issues and PRs. If interest grows beyond what I can handle solo, I'm open to co-maintainers.
https://github.com/samuelurones28/aicontext-commit
