multi-model-meta-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data in the form of analysis or feedback from other AI models (e.g., Claude, GPT, Gemini). This data is treated as a primary input for synthesis and could potentially contain instructions meant to deceive the agent. However, the skill provides explicit instructions to verify all claims against the actual source code, which serves as a significant security control.
- Ingestion points: User-provided feedback and analysis from multiple LLMs as described in the SKILL.md process.
- Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands within the model outputs being analyzed.
- Capability inventory: The skill instructs the agent to use file system tools (Grep, Glob, and Read) to examine the codebase.
- Sanitization: The skill implements a robust verification logic where claims must be confirmed or refuted by examining the actual source code before inclusion in the final report.
Audit Metadata