clinpgx-database
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches pharmacogenomics data including gene metadata, drug information, and clinical guidelines from the ClinPGx REST API at api.clinpgx.org. The implementation follows development best practices by using a dedicated session object and enforcing rate limits.
- [COMMAND_EXECUTION]: The helper script scripts/query_clinpgx.py provides utility functions to export processed medical data to local CSV files using the pandas library. It also includes instructions for manual compilation using the py_compile module for verification.
- [INDIRECT_PROMPT_INJECTION]: The skill interacts with an external data source, creating a standard attack surface for data-driven agent skills.
- Ingestion points: External medical data is ingested through the safe_api_call function in scripts/query_clinpgx.py via the ClinPGx API endpoints.
- Boundary markers: The skill does not implement specific delimiters or instructions to ignore embedded prompts within the retrieved pharmaceutical data.
- Capability inventory: The skill possesses network GET capabilities for API access and the ability to write to the local filesystem for caching and data export.
- Sanitization: Data is parsed as structured JSON and handled via pandas DataFrames; no secondary sanitization or instruction-filtering is applied to the data payload text.
Audit Metadata