commit-context

To write git commit messages that explain the why, you can generate story-driven Conventional Commits narratives directly from your current conversation context. This approach reads your code diffs and builds a commit body focused on task intent and rationale for future readers.

A module registry tracks decisions by writing MODULE-tagged blocks keyed to a `.claude/MODULES.md` file. This captures structured decision history during your session, ensuring consistent downstream distillation of why specific code changes were made.

Yes, you can include Claude Code token usage in your commit messages by pulling session cost context via the `ccusage` dependency. The skill automatically appends this usage data to the final Conventional Commits formatted output when available.

No, you do not need to manually create a module registry. The process can optionally bootstrap the `.claude/MODULES.md` registry and `CLAUDE.md` file once, with your explicit approval, if they are not already present in your repository.

The best way to automate Conventional Commits formatting is to derive the commit narrative from actual interactive development session context. It analyzes both staged and unstaged diffs to output a summary plus a short body explaining the motivation behind the changes.

You should use conversation context because diffs only show what changed, while session context provides the motivation. Deriving the commit narrative from the actual conversation prevents vague messages and helps future readers understand the intent behind the code.