web-research
Pass
Audited by Gen Agent Trust Hub on Feb 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local JavaScript scripts (
search.jsandcontent.js) located in~/agent-tools/brave-search/via the system shell to perform web searches and markdown content extraction. - [PROMPT_INJECTION]: An indirect prompt injection surface is present due to the ingestion of arbitrary web content. Ingestion points: Search result snippets and full-page content retrieved from the web via
search.jsandcontent.js. Boundary markers: The skill documentation suggests a delimited output format using textual headers (e.g., '--- Result 1 ---'). Capability inventory: The agent can execute shell-based search tools and process the resulting data to inform its responses. Sanitization: No explicit sanitization or filtering of external content is mentioned in the skill definition.
Audit Metadata