brave-search
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection attacks by fetching and presenting content from arbitrary external websites to the AI agent.
- Ingestion points: The scripts
content.jsandsearch.jsfetch HTML content from URLs supplied as arguments or retrieved from search results. - Boundary markers: The extracted markdown content is returned to the agent without clear delimiters or instructions to ignore potential commands embedded within the text.
- Capability inventory: The skill has the capability to make network requests and output text to the agent's context.
- Sanitization: The skill performs structural sanitization by converting HTML to Markdown, but it does not filter or sanitize the actual text for natural language instructions directed at the agent.
Audit Metadata