web-search
Pass
Audited by Gen Agent Trust Hub on Aug 6, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches configuration and installation documentation from the platform's GitHub repository at inference-sh/skills.
- [COMMAND_EXECUTION]: Relies on the 'belt' CLI to perform web searches and AI-driven content analysis, consistent with the tool's intended use and the skill's configuration constraints.
- [PROMPT_INJECTION]: The skill facilitates RAG workflows that process external web data, creating an indirect prompt injection surface.
- Ingestion points: Content is retrieved from the web using Tavily and Exa search applications (SKILL.md).
- Boundary markers: No specific delimiters or instruction-ignore warnings are present in the example workflows.
- Capability inventory: Processes ingested data via 'belt app run' using various AI models across search and extraction tasks.
- Sanitization: No validation or sanitization of external content is documented in the skill's research and summary examples.
Audit Metadata