alterlab-chembl
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill retrieves scientific data from the ChEMBL API, which constitutes a potential ingestion surface for indirect prompt injection. However, because ChEMBL is a professionally curated database from a well-known scientific organization (EBI), the risk is negligible.
- Ingestion points: Bioactivity measurements, molecule properties, and target information retrieved via the
chembl_webresource_clientinscripts/example_queries.pyand referenced inSKILL.md. - Boundary markers: None identified in the provided example scripts; data is processed as structured dictionaries/dataframes.
- Capability inventory: The skill is authorized to use
Bashfor package management (uv pip) and running query scripts (uv run). - Sanitization: Results are returned directly from the ChEMBL client to the agent context without additional filtering for natural language instructions.
Audit Metadata