The Agent Skills Directory

[SAFE]: The skill is a specialized security auditor for static code analysis. Its instructions and reference files establish a clear security-focused persona following industry standards like OWASP Top 10:2025 and MASVS.
[PROMPT_INJECTION]: Automated detection of prompt injection patterns in reference/ai-security.md and reference/vulnerability-patterns.md are false positives. These files contain regex patterns and heuristics specifically for the agent to use when detecting injection attacks in other code, rather than attempting to override its own instructions.
[DATA_EXPOSURE]: The skill includes extensive patterns for detecting secrets and PII. No hardcoded credentials or unauthorized data exfiltration paths were found within the skill's own assets.
[REMOTE_CODE_EXECUTION]: While the skill references numerous external security tools (Semgrep, CodeQL, MobSF, Trivy), it does so for documentation and integration purposes. It does not contain instructions to download or execute unverified third-party scripts.
[SUPPLY_CHAIN]: The skill includes detailed documentation on supply chain threats (e.g., slopsquatting, typosquatting) to assist the auditor in identifying these risks in user projects.

sentinel