The Agent Skills Directory

[SAFE]: The skill focuses on read-only inspection of the local environment using standard Git and utility commands (git status, git diff, git log, grep). No network operations or external data exfiltration patterns were detected.- [SAFE]: Instructions explicitly prohibit destructive Git commands such as reset, checkout, clean, or stash, ensuring the integrity of the user's worktree is maintained.- [SAFE]: The skill encourages a 'smallest evidence surface' approach, directing the agent to use scoped, path-bounded commands rather than dumping entire repository histories or large diffs into the context.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from Git history and diffs. While this is a known attack surface, the risk is minimal in this context as the skill is designed for information gathering and does not chain this data into high-risk operations like code execution or network requests.

pragmatic-git-context