tooluniverse-kegg-disease-drug
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is a legitimate research tool with no evidence of prompt injection, data exfiltration, or persistence mechanisms. The defined tools and workflows align with the stated purpose of KEGG-based biological research.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external biological databases (KEGG), which represents a surface for indirect prompt injection. 1. Ingestion points: External database content enters the agent context via tools such as KEGG_get_disease, KEGG_search_drug, and KEGG_get_variant (SKILL.md). 2. Boundary markers: Absent. The skill instructions do not specify explicit delimiters or warnings for the agent to ignore instructions embedded within retrieved biological data. 3. Capability inventory: The skill is limited to read-only data retrieval and identifier conversion. It lacks dangerous capabilities such as file system writing, outbound network POST operations, or arbitrary command execution across all scripts. 4. Sanitization: Absent. There is no explicit evidence of sanitization or filtering of the content returned from the external API within the provided skill instructions.
Audit Metadata