The Agent Skills Directory

[SAFE]: The skill instructions explicitly enforce a read-only boundary, delegating all repository mutations to separate delivery tools, which prevents accidental or malicious code modification during the audit process.
[SAFE]: The inclusion of the validate-frontend-visual-evidence.py script provides a local, offline mechanism for validating structured audit data. The script uses only Python standard libraries and includes specific security checks, such as verifying that resource locators in the audit JSON do not contain path traversal patterns (e.g., ../).
[SAFE]: External documentation and references point exclusively to reputable open-source projects and official framework documentation on GitHub, serving as legitimate architectural exemplars without introducing untrusted code execution or third-party dependencies.

audit-frontend