skill-developer
Pass
Audited by Gen Agent Trust Hub on Feb 19, 2026
Risk Level: SAFENO_CODE
Full Analysis
- SAFE (SAFE): The provided files (ADVANCED.md, HOOK_MECHANISMS.md, PATTERNS_LIBRARY.md, SKILL_RULES_REFERENCE.md, TRIGGER_TYPES.md) are entirely documentation. They provide architectural overviews, configuration schemas, and pattern libraries for a skill management system.
- NO_CODE (SAFE): No executable scripts (.sh, .ts, .py) or active configuration files (skill-rules.json) were included in the skill payload. The files contain examples and instructions for developers to build their own skills but do not perform any actions themselves.
- Data Exposure & Exfiltration (SAFE): The documentation mentions system variables like session_id and transcript paths for hook inputs, but this is within the context of standard agent-tool communication. No hardcoded credentials or exfiltration logic was detected.
- Indirect Prompt Injection (SAFE): While the described system (hooks processing user prompts) constitutes a surface for indirect prompt injection, the documentation itself focuses on using these hooks to enforce safety guardrails and domain-specific best practices rather than bypassing them.
Audit Metadata