diffdock
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [SAFE] (SAFE): Extensive review of the scripts and configuration files reveals no malicious patterns, hardcoded secrets, or unauthorized network operations.
- [INDIRECT_PROMPT_INJECTION] (LOW): The script
scripts/prepare_batch_csv.pyrepresents a data ingestion surface as it processes user-provided CSV files. Evidence Chain: 1. Ingestion points: CSV file path passed via command line. 2. Boundary markers: No explicit prompt boundary markers; however, data is treated as structured CSV content. 3. Capability inventory: Script performs file existence checks and writes validated CSV output. 4. Sanitization: Includes SMILES string validation using RDKit and verification of required column headers. - [DYNAMIC_EXECUTION] (LOW): The script
scripts/setup_check.pyutilizes dynamic loading via__import__to verify the presence of scientific dependencies. This is categorized as LOW risk because it uses a predefined list of trusted package names and does not incorporate untrusted external input into the import call.
Audit Metadata