The Agent Skills Directory

[SAFE]: The skill instructions and documentation focus on improving code quality and security without introducing malicious logic.- [SAFE]: No credential exposure, sensitive file access, or unauthorized network communication was detected.- [SAFE]: No remote code execution, persistence mechanisms, or privilege escalation patterns were found.- [SAFE]: Indirect Prompt Injection: The skill reviews user-provided code (Ingestion Point: diff input). Boundary markers are absent. Capability inventory includes no subprocess or network operations. Sanitization is absent. The risk is evaluated as safe given the lack of exploitable tools.

37signals-review