The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard Git commands (git status, git diff, git add, git commit) to manage the local repository as requested by the user. These commands are fundamental to the skill's purpose and are executed locally.
[REMOTE_CODE_EXECUTION]: Commit message formatting is handled by internal scripts (scripts/compose_commit_message and its Python counterpart). No external code is downloaded or executed, and the system relies on pre-packaged tools.
[DATA_EXFILTRATION]: There are no network operations or instructions to send data externally. The skill interacts only with the local filesystem and the Git command-line interface.
[PROMPT_INJECTION]: While the skill processes untrusted repository data via git diff and git status (ingestion points in SKILL.md), it includes robust boundary markers such as a mandatory confirmation checkpoint before commit execution (SKILL.md Section 5). The capability inventory includes local file writes and Git commands, but risks are mitigated by Python-based message normalization in scripts/compose_commit_message.py and the requirement for explicit user approval of generated text.
[SAFE]: The workflow incorporates security best practices, including a 'Preflight' inspection of changes and a mandatory confirmation step before the final commit action. This prevents unintended modifications to the repository.

git-commit