tooluniverse-adverse-outcome-pathway
Pass
Audited by Gen Agent Trust Hub on Mar 27, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a set of instructions for retrieving and analyzing toxicological data from reputable scientific sources (PubChem, AOPWiki, and CTD). It does not contain executable code, credential access, or network calls to untrusted domains.- [PROMPT_INJECTION]: The skill interacts with external data from scientific databases. While this creates a surface for indirect prompt injection, the risk is mitigated by the fact that the agent is restricted to a set of read-only tools and does not have access to sensitive system functions or write operations.
- Ingestion points: External database results from AOPWiki, PubChemTox, and CTD tools.
- Boundary markers: None explicitly defined in the instructions.
- Capability inventory: No file writing, command execution, or arbitrary network access.
- Sanitization: Not applicable for the information-gathering scope of this skill.
Audit Metadata