alterlab-rowan
Pass
Audited by Gen Agent Trust Hub on Apr 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides legitimate integration with the Rowan computational chemistry platform, documenting scientific workflows like molecular property prediction and docking.
- [CREDENTIALS_UNSAFE]: The skill correctly handles authentication by recommending environment variables and using safe documentation placeholders for API keys.
- [EXTERNAL_DOWNLOADS]: The skill installs the official 'rowan-python' package and communicates with authorized vendor domains like 'labs.rowansci.com' for calculation submissions.
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it processes external scientific data. * Ingestion points: Reads SMILES strings, protein sequences, and structural files in SKILL.md and references/proteins_and_organization.md. * Boundary markers: No specific delimiters are used to wrap processed data strings. * Capability inventory: Performs network API calls (rowan.submit_) and file system downloads (workflow.download_) as documented in SKILL.md and references/api_reference.md. * Sanitization: No explicit input sanitization is documented in the skill instructions.
Audit Metadata