legal-industry-report
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes untrusted regulatory and industry content from the web to generate reports, establishing a surface for potential indirect prompt injection.
- Ingestion points: Dynamically fetches legal data and industry rules from external URLs as defined in the registry files.
- Boundary markers: Includes mandatory lawyer review steps and explicit research disclaimer requirements before report finalization.
- Capability inventory: The skill spawns local processes for script execution, utilizes browser-based rendering via Playwright, and performs local file system writes.
- Sanitization: Employs Jinja2 templates with auto-escaping and includes automated structural gatekeeping via validation scripts.
- [EXTERNAL_DOWNLOADS]: Fetches regulatory information and policy updates from well-known official government and statutory disclosure domains (e.g., gov.cn, cninfo.com.cn) to inform its research process.
- [COMMAND_EXECUTION]: Spawns local subprocesses to execute internal orchestration scripts (validation, rendering, PDF generation) and optional diagram rendering tools. These operations use list-based argument passing to prevent shell injection vulnerabilities.
Audit Metadata