drugbank-database
Pass
Audited by Gen Agent Trust Hub on Feb 23, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill uses 'pickle.load()' to read cached data in 'references/data-access.md', which is an insecure practice that could allow arbitrary code execution if the local cache file is compromised.
- [PROMPT_INJECTION]: The skill processes untrusted external data from the DrugBank database, creating a surface for indirect prompt injection.
- Ingestion points: Detailed in 'references/data-access.md', the skill downloads a comprehensive XML database from 'go.drugbank.com'.
- Boundary markers: No explicit instructions or markers are used to separate database content from agent instructions.
- Capability inventory: The skill has file system write access for caching and network access for downloads, as seen in 'references/data-access.md' and 'scripts/drugbank_helper.py'.
- Sanitization: The skill lacks explicit sanitization of text data extracted from the database.
Audit Metadata