CVE-2026-91943: Crawl4AI before 0.9.3 SSRF via PDFContentScrapingStrategy
Crawl4AI before 0.9.3 contains a server-side request forgery vulnerability in PDFContentScrapingStrategy where getpdfpath() re-downloads targets with Python requests without egress validation. Authenticated attackers can supply URLs that redirect to internal addresses or use DNS rebinding to access internal services, exfiltrating responses through PDF text extraction in crawl results.
Affected Software
Event History
Frequently Asked Questions
Who is exposed to this issue?
Deployments of Crawl4AI earlier than 0.9.3 are exposed where authenticated users can supply URLs processed by PDFContentScrapingStrategy. The vulnerable server must be able to make requests to the targeted internal address or service.
What does an attacker need to exploit it?
An attacker needs authenticated access and the ability to provide a URL for PDF processing. They can use redirects to internal addresses or DNS rebinding, and retrieve internal service responses through PDF text extraction in crawl results.
Is user interaction required?
No. The vulnerability has no user-interaction requirement; an authenticated attacker can trigger the server-side request through the affected processing path.
What should be done if patching cannot happen immediately?
The provided data does not specify a workaround. Prioritize restricting authenticated users' ability to submit untrusted URLs to PDFContentScrapingStrategy and limiting the crawler's network access to internal services where possible.