Executive brief
gpt-crawler is a web scraping tool that creates knowledge bases from website content. A vulnerability in the file output feature allows attackers to write arbitrary files anywhere on the server's filesystem by manipulating the output filename parameter, potentially overwriting critical system or application files.
Technical details
The POST /crawl endpoint in gpt-crawler through version 1.5.1 fails to properly validate or sanitize the outputFileName parameter, allowing path traversal attacks. An unauthenticated attacker can supply absolute paths or parent-directory traversal sequences (e.g., "../../../etc/passwd") to write files to arbitrary locations on the filesystem. The vulnerability is exploitable remotely without authentication, and an attacker can combine this with attacker-controlled URL content to write malicious files to sensitive directories, such as web root or application configuration directories, leading to code execution or data destruction.
Affected products
- BuilderIO gpt-crawler through 1.5.1
Timeline
- 2026-08-28: disclosed