firecrawl
Pass
Audited by Gen Agent Trust Hub on Apr 4, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted web content retrieved via the Firecrawl API, creating a surface for indirect prompt injection. \n
- Ingestion points:
scripts/scrape.py,scripts/crawl.py, andscripts/search.pyfetch data from arbitrary external URLs. \n - Boundary markers: The scripts print retrieved content (markdown/HTML) directly to the output without using clear delimiters to isolate it from the agent's instructions. \n
- Capability inventory: The skill uses
urllib.requestto perform network requests to the Firecrawl API. \n - Sanitization: Content is displayed without sanitization, allowing potential instructions in the scraped data to reach the agent.
Audit Metadata