The Agent Skills Directory

[SAFE]: The skill performs a structured analysis of local documentation files within a project's repository. It strictly follows a read-only policy and does not attempt to modify any files or exfiltrate data.
[SAFE]: The auto-discovery mechanism uses predefined, safe paths (e.g., docs/strategy/, VISION.md) and explicitly excludes dependency folders like node_modules and .venv, preventing accidental processing of untrusted third-party code.
[SAFE]: No obfuscation, prompt injection attempts, or remote code execution patterns were detected in the skill's instructions or metadata.

project-review