create-life-sciences-patent-report-ls
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill includes comprehensive instructions to prevent data leakage. In 'references/report-spec.md', it mandates a 'static route audit' to check for API keys, authorization tokens, and absolute local developer paths within the generated HTML. It also requires the report to be self-contained by prohibiting external dependencies like CDNs or remote scripts, which prevents unauthorized network exfiltration or tracking.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from various external MCP sources including patent searches ('advanced_patent_search'), clinical trial registries ('clinical_trials'), and news feeds ('current_awareness').
- Ingestion points: Data retrieved from external pharmaceutical and legal databases via MCP tools.
- Boundary markers: Uses a structured 'source record' JSON schema and a 'source-link rule' to maintain data lineage.
- Capability inventory: Capabilities are limited to generating a self-contained HTML report.
- Sanitization: The skill incorporates a 'Validation protocol' that includes HTML structure parsing and a 'Link audit table' to verify identifiers against the source map.
Audit Metadata