Executive brief
Firecrawl, an open-source tool used to scrape and convert websites into data for AI models, contains a security flaw in its web scraping component. An attacker can trick the system into accessing private internal company servers or sensitive data by using a malicious web link that redirects to a restricted internal address. This could lead to the exposure of sensitive internal information or unauthorized access to private network services.
Technical details
A Server-Side Request Forgery (SSRF) vulnerability exists in the Playwright scraping service of Firecrawl due to insufficient validation of HTTP redirects. While the service validates the initial user-provided URL against network policies, it fails to re-validate the destination URL if the initial request returns a redirect (e.g., 301/302 status codes). An unauthenticated remote attacker can exploit this by hosting a malicious URL that redirects to internal or restricted IP addresses (such as link-local or private network ranges). This allows the attacker to bypass SSRF protections and exfiltrate data from internal services. The vendor recommends using a secure proxy server (via PROXY_SERVER environment variable) configured to block link-local and private traffic as a primary mitigation.
Affected products
- Firecrawl Firecrawl <= 2.8.0
Timeline
- 2026-03-26: advisory: Initial disclosure by VulnCheck
- 2026-03-26: disclosed