The Agent Skills Directory

[SAFE]: The skill is a configuration utility for Gitleaks. It provides templates (gitleaks.toml) and bash scripts (sanitize-check.sh, test_gate.sh) designed to scan local repository files for secrets.
[COMMAND_EXECUTION]: The skill includes shell scripts that execute standard developer tools (git, grep, gitleaks, python3). These operations are local to the repository where the skill is applied and are consistent with the skill's stated purpose of auditing code for leaks.
[CREDENTIALS_UNSAFE]: While the skill handles credentials, it does so to prevent their leakage. It explicitly instructs users to store real secrets in .env files (which it adds to the gitleaks allowlist to avoid self-matching) and provides placeholder patterns in its templates rather than hardcoding real secrets.
[INDIRECT_PROMPT_INJECTION]: The skill involves processing file content, but it uses external deterministic tools (grep, gitleaks) rather than passing raw file content directly into an LLM context for evaluation, significantly reducing the surface area for this attack category.

setup-gitleaks