The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses git commands like git filter-repo and bfg to perform destructive history rewriting, which is the intended purpose for cleaning leaked credentials. It also generates a standard pre-commit hook bash script and makes it executable with chmod +x.
[EXTERNAL_DOWNLOADS]: The guide recommends installing established security tools such as git-filter-repo via pip and git-secrets via Homebrew.
[DATA_EXFILTRATION]: The skill searches for sensitive files like .env, cloud credentials, and SSH keys within the repository history. This data exposure is a core functional requirement for secret scanning and no external exfiltration was detected.
[PROMPT_INJECTION]: The skill processes file names and contents from the repository, creating an indirect prompt injection surface.
Ingestion points: Untrusted data enters through git log, find, and git grep in references/full-guide.md.
Boundary markers: None provided in the command output.
Capability inventory: Includes package installation, history rewriting, and force pushing to remotes.
Sanitization: None performed on the captured file data.

git-safety