The Agent Skills Directory

[SAFE]: The skill adheres to the principle of least privilege by performing only standard Git operations necessary for its stated purpose.
[SAFE]: It incorporates a mandatory human-in-the-loop confirmation step (Step 4 and Step 7) before executing any state-changing commands like git commit.
[SAFE]: Security-conscious instructions are included to proactively scan for and avoid committing sensitive files such as .env files or hardcoded credentials (Step 3).
[PROMPT_INJECTION]: The skill processes untrusted data from local repository history and diffs, representing an indirect prompt injection surface.
Ingestion points: Git status, diff, and log outputs are processed in Step 1, 2, and 3.
Boundary markers: The skill uses clear markdown headers and distinct workflow steps to maintain instruction context.
Capability inventory: The agent can execute git add and git commit commands via the shell.
Sanitization: The skill relies on explicit user review of proposed commit messages and strategies before any commands are run, providing a manual safeguard against malicious content in diffs.

git-committer