What problem does it solve? It turns the changes made during a coding session into clean, well-organized git commits without requiring the user to manually review diffs, group files, and write commit messages from scratch. ## Core Features & Use Cases - Session-Aware Commit Planning: Reviews conversation history, git status, and git diff to decide whether changes should be one commit or multiple logical commits. - Approval-Gated Execution: Presents the planned files and commit messages to the user and only commits after explicit confirmation. - Clean Attribution: Writes commit messages in imperative mood as if authored by the user, with no co-author lines or AI attribution. - Use Case: After a long refactoring session, ask the agent to commit the work; it groups related files into atomic commits, proposes messages, and executes them once you approve. ## Quick Start Ask the agent to commit the changes from this session and approve the proposed commit plan when it is presented.