genomic-intelligence
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill facilitates communication with
api.genomicintelligence.aifor model inference andrest.ensembl.orgfor reference genome retrieval. - [INDIRECT_PROMPT_INJECTION]: The skill ingests genomic data, which represents an attack surface for data-driven instruction injection.
- Ingestion points: Sequence data enters through the
sequenceREST body field and MCP tool handles. - Boundary markers: Structured API schemas and handle-based data management keep sequence content distinct from the agent's logic flow.
- Capability inventory: Access is restricted to network operations via the
requestslibrary; there is no local code execution or file system access. - Sanitization: Input sequences are validated for biological validity and cleaned of whitespace at the server level.
Audit Metadata