news-search
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWPROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection (LOW): The skill retrieves content from external news articles which can contain malicious instructions designed to manipulate the agent's subsequent reasoning or actions.
- Ingestion points: The
results[].title,results[].description, andresults[].extra_snippetsfields described in the response format. - Boundary markers: None specified; the skill returns raw strings from external sources.
- Capability inventory: The skill is a data retrieval tool (read-only) and lacks capabilities for file system modification, command execution, or outbound network requests beyond the Brave API itself.
- Sanitization: No sanitization or filtering of the retrieved web content is performed by the skill before it is provided to the agent.
Audit Metadata