The Agent Skills Directory

[COMMAND_EXECUTION]: The skill heavily utilizes shell commands, specifically git, grep, and chmod, to perform its core functions. It defines safe workflows for destructive operations such as git push --force and git filter-repo by mandating dry runs, backups, and user confirmation.
[EXTERNAL_DOWNLOADS]: The instructions guide the user or agent to install widely-used security tools including pre-commit, git-secrets, and git-filter-repo via official package managers (pip and Homebrew) and GitHub. These are well-known services and the tools are standard in the security ecosystem.
[PERSISTENCE]: The skill establishes persistence within the local repository by installing a pre-commit hook script in .git/hooks/pre-commit. This script is designed to run automatically on every commit to enforce security checks, which is the intended and transparent purpose of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an ingestion surface for untrusted data as it processes git diff and git status outputs to identify potential secrets.
Ingestion points: git diff --cached and git status --porcelain in SKILL.md and references/full-guide.md.
Boundary markers: The skill uses narrow regex patterns for detection and instructs the agent to report findings rather than execute them directly.
Capability inventory: The skill has the ability to modify repository hooks, write to the filesystem, and push to remote repositories.
Sanitization: The logic relies on fixed regular expressions to filter data, reducing the risk of the agent interpreting malicious data as instructions.

git-safety