tooluniverse-scientific-peer-review
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external content such as manuscripts, protocols, and data, creating an indirect prompt injection surface.
- [INGESTION_POINTS]: User-supplied manuscripts, supplementary files, and quoted text are processed by the agent.
- [BOUNDARY_MARKERS]: The Trust Boundary section in SKILL.md provides explicit instructions to treat inputs as data, not instructions, and to disregard any embedded commands such as requests to ignore the skill or predetermine ratings.
- [CAPABILITY_INVENTORY]: The skill does not request or use any dangerous tools (e.g., shell execution, file system modification, or non-whitelisted network access), preventing the exploitation of the injection surface.
- [SANITIZATION]: Instructions mandate grounding all reviews in provided evidence and labeling unverified external claims appropriately.
- [SAFE]: No malicious code, obfuscation, or unauthorized data access patterns were found in the skill's instructions or referenced documents.
Audit Metadata