company-profiling
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface through its integration with external MCP tools and web search results, creating a theoretical vector for indirect prompt injection.
- Ingestion points: Untrusted data enters the agent context via MCP database fetches and web search tool outputs (SKILL.md).
- Boundary markers: The prompt instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings when processing fetched results.
- Capability inventory: The skill is scoped to pharmaceutical analysis and lacks high-risk capabilities like file system modifications or unauthorized network requests.
- Sanitization: No specific filtering or sanitization steps are defined for the ingested pharmaceutical or web data before processing.
Audit Metadata