The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes common security auditing tools such as trufflehog, gitleaks, and git-dumper. These tools are used according to their intended purpose for identifying hardcoded secrets and security misconfigurations in repositories.
[INDIRECT_PROMPT_INJECTION]: The skill involves processing external data sources, including git repositories and CI/CD configurations. This ingestion surface is considered safe within the context of the skill's primary purpose as a security auditing tool, particularly as it includes explicit warnings against data exfiltration and provides best practices for validating found credentials.
[DATA_EXPOSURE]: While the skill deals with discovering sensitive data (secrets), it provides safe handling instructions, advising the agent to use read-only validation methods and to avoid pasting actual secret values into reports.

code-review-secrets-detection