The Agent Skills Directory

[SAFE]: The skill provides legitimate instructional guidelines for reviewing the quality and completeness of knowledge artifacts. The use of emphasis and repetitive directives (e.g., "MANDATORY IMPORTANT MUST ATTENTION") is intended to ensure task thoroughness and adherence to project conventions like DRY (Don't Repeat Yourself) and OOP (Object-Oriented Programming), rather than to bypass safety protocols.
[SAFE]: The skill implements a multi-round review requirement (Round 1 for understanding and Round 2 for catching missed issues). This is a positive safety and quality pattern that forces the agent to re-evaluate its initial findings and reduces the risk of oversight or hallucinations.
[SAFE]: The skill is designed to process external "knowledge artifacts" using tools like Read, Grep, and Bash. While this creates a surface for indirect prompt injection, the skill is intended for auditing and analysis, and no evidence of malicious intent or exploitative patterns was found. The tool usage is constrained to the filesystem and matches the declared configuration.
[SAFE]: No obfuscation, hardcoded credentials, unauthorized network operations, or persistence mechanisms were detected in the instructions or metadata.

knowledge-review