varied-search
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process content from external, untrusted sources including the web, GitHub, and X/Twitter.\n
- Ingestion points: Data returned from 5x web searches, 5x GitHub scrapes, and 5x Twitter scrapes via the DeepAPI tool (SKILL.md).\n
- Boundary markers: No boundary markers or "ignore instructions" directives are present to prevent the agent from following malicious instructions hidden in the scraped content.\n
- Capability inventory: The skill generates a Markdown report based on this external data. While no dangerous capabilities like file writing or command execution are explicitly invoked in the instructions, the agent's general capabilities are exposed to potentially poisoned content.\n
- Sanitization: No sanitization or filtering of the external content is performed before interpolation into the final report.
Audit Metadata