alphagenome
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security issues were detected. The skill uses environment variables for sensitive API keys and implements clear validation logic for variant strings and genomic intervals.\n- [EXTERNAL_DOWNLOADS]: The skill downloads genomic annotations from a Google Cloud Storage bucket (storage.googleapis.com) and interacts with the AlphaGenome API on Google-owned infrastructure (gdmscience.googleapis.com). These are recognized as trusted and well-known services.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external genomic data files (VCF, TSV, CSV). It mitigates injection risks by strictly validating input using regular expressions for standard genomic coordinate formats and restricting data usage to structured API parameters and local processing scripts.\n
- Ingestion points: atlas_query.py and score_variants.py ingest user-provided files via --input flags.\n
- Boundary markers: The parsing logic enforces strict patterns (e.g., chr:pos:ref>alt), effectively creating a boundary between valid data and arbitrary text.\n
- Capability inventory: The skill performs gRPC network requests to the AlphaGenome API and writes results to local TSV/CSV/JSON/Parquet files.\n
- Sanitization: All inputs are filtered through normalization functions and regex checks in _common.py before being processed.
Audit Metadata