deepchem
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection because it ingests data from external sources and maintains access to powerful tools.\n
- Ingestion points: Molecular data is loaded from external files using specialized loaders such as
CSVLoader,SDFLoader,FASTALoader, andJsonLoader, as documented inSKILL.mdandreferences/api_reference.md.\n - Boundary markers: There are no instructions or boundary markers to differentiate between data and instructions within the processed chemical files, nor are there warnings for the agent to ignore potential instructions embedded in the data.\n
- Capability inventory: The skill is configured with access to
Bash,Read,Write, andEdittools. Additionally, it provides several Python scripts (scripts/graph_neural_network.py,scripts/predict_solubility.py,scripts/transfer_learning.py) that perform complex operations including model training and evaluation.\n - Sanitization: The skill does not implement validation or sanitization of the input files (SMILES strings, protein sequences, etc.) before they are used in data processing or model training workflows.
Audit Metadata