diffdock
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- Prompt Injection (SAFE): No instructions attempting to bypass safety filters or override agent behavior were found. All markdown content is technical documentation for molecular docking.
- Data Exposure & Exfiltration (SAFE): No hardcoded credentials, sensitive file paths, or unauthorized network operations were detected. The scripts interact with local CSV and chemical files (SDF, PDB) as expected for the domain.
- Obfuscation (SAFE): All scripts and configuration files are written in clear, readable text. No Base64, zero-width characters, or homoglyph-based evasion techniques were identified.
- Unverifiable Dependencies & Remote Code Execution (SAFE): The
setup_check.pyscript identifies necessary dependencies but does not automatically download or execute them. It provides standard installation instructions for well-known scientific libraries (PyTorch, RDKit, etc.). - Privilege Escalation (SAFE): No use of
sudo,chmod 777, or other privilege-altering commands was found. - Persistence Mechanisms (SAFE): The skill does not attempt to modify shell profiles, cron jobs, or registry keys.
- Indirect Prompt Injection (SAFE): While the skill processes external chemical data (SMILES, CSV), the utility scripts perform structural validation (using RDKit) and do not interpolate untrusted data into an agent prompt context in an exploitable manner.
- Dynamic Execution (SAFE): The
setup_check.pyscript uses__import__to verify the presence of locally installed packages from a static list, which is a standard diagnostic practice and not an attack vector.
Audit Metadata