brenda-database
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the external BRENDA database, which could contain malicious instructions if the database content were compromised.
- Ingestion points: Data is fetched from the BRENDA SOAP API in scripts/brenda_queries.py (e.g., get_km_values, get_reactions).
- Boundary markers: There are no explicit boundary markers or instructions to the agent to disregard potential commands embedded within the retrieved biochemical data.
- Capability inventory: The skill includes file writing capabilities for exporting data and saving visualizations (scripts/brenda_queries.py, scripts/brenda_visualization.py).
- Sanitization: The retrieved data, including free-text commentary fields, is parsed using regular expressions but is not sanitized for potential injection patterns before being returned to the agent context.
- [EXTERNAL_DOWNLOADS]: The skill connects to a remote service to fulfill its primary function.
- Official Source: Fetches enzyme data from the well-known BRENDA (Braunschweig Enzyme Database) service at brenda-enzymes.org using the SOAP protocol.
- Configuration: Requires user-provided credentials via environment variables (BRENDA_EMAIL, BRENDA_PASSWORD) to authenticate with the remote API.
Audit Metadata