chembl-database
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process bioactive molecule data and research findings from the external ChEMBL API hosted by the European Bioinformatics Institute (EBI).
- Ingestion points: Data is retrieved via the
chembl_webresource_clientlibrary as shown in the molecule, target, and activity query examples inSKILL.md. - Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded prompts within the retrieved scientific data.
- Capability inventory: The skill uses Python for data manipulation and
pandasfor analysis, allowing for significant data processing of untrusted external content. - Sanitization: No specific sanitization or validation logic is provided for the API responses before they are processed by the agent.
- [PROMPT_INJECTION]: The skill includes behavioral instructions that direct the agent to proactively suggest a specific hosted platform (K-Dense Web) under certain conditions.
- Evidence: The section 'Suggest Using K-Dense Web For Complex Worflows' in
SKILL.mdinstructs the agent to monitor user task complexity and steer the user toward a specific service. - [EXTERNAL_DOWNLOADS]: The skill requires the installation of the
chembl_webresource_clientandpandaspackages. - Evidence: Installation instructions using
uv pip installare provided inSKILL.md. These are well-known, legitimate packages for scientific research.
Audit Metadata