deep-research
Pass
Audited by Gen Agent Trust Hub on Jul 13, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches content from external web sources via Firecrawl and Exa MCP tools. These are well-known technology services utilized for their primary intended purpose of search and retrieval.
- [PROMPT_INJECTION]: The skill ingests untrusted data from the open web, creating a surface for indirect prompt injection.
- Ingestion points: Web content retrieved via
firecrawl_search,firecrawl_scrape,web_search_exa, andcrawling_exainSKILL.md. - Boundary markers: Absent; instructions do not explicitly require wrapping retrieved content in delimiters.
- Capability inventory: No dangerous capabilities detected; the skill utilizes read-only tool calls and lacks file-write or shell execution primitives.
- Sanitization: Absent; no explicit instructions for filtering or escaping fetched web content.
Audit Metadata