narrative-cascade-planner
Pass
Audited by Gen Agent Trust Hub on Aug 20, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external content through web scraping and user pastes, creating a potential injection surface. Ingestion points: Scraped pages (via firecrawl.py and wayback.py) and pasted surface exports. Boundary markers: References SECURITY.md for handling untrusted input. Capability inventory: Execution of internal scripts (registry-events.py) and writing to local memory directories. Sanitization: Explicitly instructs the agent to never follow instructions embedded in external content.\n- [COMMAND_EXECUTION]: The skill utilizes local scripts for operations. Evidence: Invokes firecrawl.py, wayback.py, and registry-events.py. These scripts are utilized for their intended purposes of surface scraping and data registry updates.\n- [DATA_EXPOSURE]: The skill interacts with local project data. Evidence: Reads and writes to directories under the memory/ path. This data access is restricted to narrative and marketing claims data relevant to the skill's purpose.
Audit Metadata