competitor-teardown
Pass
Audited by Gen Agent Trust Hub on Aug 3, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: References external documentation from the inference-sh GitHub repository and instructions for installing the belt-sh CLI tool via npx.
- [COMMAND_EXECUTION]: Utilizes the belt CLI to run various specialized agents for search, browsing, and image processing, limited by the allowed-tools configuration.
- [DYNAMIC_EXECUTION]: Provides a static Python script template for generating competitive positioning maps using matplotlib.
- [INDIRECT_PROMPT_INJECTION]: The skill processes research data from third-party websites and search engines. 1. Ingestion points: Tavily search results, Exa search results, and website extractions from agent-browser. 2. Boundary markers: Data is encapsulated within structured JSON inputs for the belt tools. 3. Capability inventory: Restricted to defined belt tool commands. 4. Sanitization: Relies on the underlying research tools to handle external content safely.
Audit Metadata