The Agent Skills Directory

[SAFE]: The skill performs legitimate repository management tasks and does not demonstrate any malicious behaviors or exfiltration patterns.- [COMMAND_EXECUTION]: Utilizes the Bash tool to run standard git commands like status, diff, add, and commit. These actions are restricted by a safety protocol that prevents destructive operations such as force pushing or modifying git configurations.- [DATA_EXPOSURE]: Explicitly instructs the agent to avoid committing sensitive files like .env, credentials, or private keys, ensuring that local secrets remain protected.- [PROMPT_INJECTION]: Ingests local diff data to generate commit messages. (1) Ingestion points: git diff and git diff --staged in SKILL.md. (2) Boundary markers: Absent. (3) Capability inventory: git add and git commit in SKILL.md. (4) Sanitization: Absent. The potential impact is limited as the processed data is used solely for generating descriptive commit text.

git-commit