deep-research
Pass
Audited by Gen Agent Trust Hub on Jul 4, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from the public web.\n
- Ingestion points: External web content is ingested via
firecrawlandexaMCP tools as described inSKILL.md(Step 3 and Step 4).\n - Boundary markers: None identified in
SKILL.md. The agent is not instructed to treat the fetched content purely as data or to ignore instructions embedded in the results.\n - Capability inventory: The skill allows for file writing (
SKILL.md, Step 6) and spawning parallel sub-agents using platform-specific task tools (SKILL.md, Parallel Research section).\n - Sanitization: None identified. The external content is summarized and synthesized directly into the final research report.
Audit Metadata