Git-reabsorb: Reorganize Git commits with new structure using an LLM (github.com)

🤖 AI Summary
git-reabsorb is a lightweight Rust-based CLI that rewrites a Git branch by unstaging and recommitting changes according to a new structure — and it can optionally use an LLM-driven strategy to decide how commits should be reorganized. Built with Rust 1.70+, the repo builds and installs with standard cargo commands (cargo build --release, cargo test, cargo install --path .). Usage is simple: run git-reabsorb for a basic rewrite or pass --strategy llm to apply the LLM-powered commit regrouping; you can also skip pre-commit hooks with --no-verify and combine flags as needed. For AI/ML projects this is notable because it automates semantic cleanup of commit history, which helps make model development, dataset changes, and experiment code more reviewable and reproducible. The LLM strategy suggests the tool reasons about the intent of changes rather than relying purely on diffs, potentially grouping related edits into coherent commits and improving changelogs. Technical constraints and practices to note: it requires Rust 1.70+, is run locally, and is dual-licensed (Apache-2.0 or MIT). Also be mindful to review any LLM-driven rewrites — privacy, hallucination or misleading commit messages are possible, and rewritten history can affect collaboration if pushed without coordination.
Loading comments...
loading comments...