The Agent Skills Directory

[SAFE]: The skill explicitly forbids the staging of sensitive files, specifically mentioning .env, credentials, and private keys. It also prohibits dangerous flags like --force or --no-verify and prevents modifications to the git configuration.
[COMMAND_EXECUTION]: The skill utilizes local shell commands (git status, git add, git commit) to manage the file system. Security is maintained through a mandatory 'STOP' step that requires the user to manually confirm the proposed commit message and action before any commit command is executed.
[PROMPT_INJECTION]: The skill analyzes file differences to generate commit messages, which presents a surface for indirect prompt injection from data within the repository files.
Ingestion points: File contents and diffs are ingested via git diff and git status (SKILL.md).
Boundary markers: No explicit delimiters are used to wrap the diff content when analyzed by the agent.
Capability inventory: The skill has the ability to stage and commit files to the local repository.
Sanitization: The risk is mitigated by the mandatory human review of the generated commit message prior to execution.

git-commit