research
Pass
Audited by Gen Agent Trust Hub on Jun 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Orchestrates search and content extraction through CLI tools like brave-search and firecrawl to facilitate deep research.
- [SAFE]: Implements secure storage of session data in XDG-compliant paths (
/.config/LEV/research) and manages API keys via local environment files (/.env.local), following security best practices. - [PROMPT_INJECTION]: Presents a surface for indirect prompt injection through web scraping operations. Ingestion points: External content fetched from search results and scrape tools like firecrawl. Boundary markers: Output structure is delimited via markdown synthesis templates. Capability inventory: Execution of local search scripts and AI-driven synthesis via oracle. Sanitization: No explicit sanitization or filtering of ingested web content is defined in the instructions.
Audit Metadata