meta-cognitive-reasoning
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is comprised of instructional content and reasoning frameworks. It lacks executable scripts, network exfiltration patterns, or unauthorized access to sensitive files.- [INDIRECT_PROMPT_INJECTION]: The skill serves to analyze external data like code and logs. 1. Ingestion points: User-supplied documentation and source code during analysis tasks (SKILL.md). 2. Boundary markers: Mandated use of explicit tool output and quoted evidence (SKILL.md). 3. Capability inventory: Instructions to utilize shell commands (grep, ls, find) for verification (SKILL.md). 4. Sanitization: Relies on evidence-based grounding rather than automated filtering.
Audit Metadata