mendelian-randomization-protocol-designer
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill package includes an evaluation report file (
eval_report_mendelian-randomization-protocol-designer_result.json) that contains self-referential safety assertions such as 'security: PASS' and 'Full marks (12/12)'. This content specifically targets safety analyzers to bias their conclusions. - Ingestion points: The file is present within the skill's file structure and processed during analysis.
- Boundary markers: No boundary markers or 'ignore' instructions are applied to this specific metadata file.
- Capability inventory: The skill is primarily instructional and does not contain code execution or network exfiltration capabilities.
- Sanitization: No sanitization is performed on the content of this evaluation report.
- [EXTERNAL_DOWNLOADS]: The skill metadata references an external repository at
github.com/aipoch/medical-research-skills. This represents a vendor-owned resource corresponding to the skill's author (AIPOCH).
Audit Metadata