Executive brief
ItzCrazyKns Vane, an AI-powered search engine, contains a critical security flaw where all of its application programming interfaces (APIs) lack authentication. This allows any remote user to access the system without a password to steal sensitive AI service API keys (such as OpenAI or Anthropic keys), read or delete private chat histories, and modify the application's core configuration. Exploitation of this vulnerability can lead to significant financial loss through stolen API credits and the total exposure of private user data.
Technical details
Vane version 1.12.1 and prior versions suffer from a complete absence of authentication and authorization mechanisms (CWE-306) across more than 16 API endpoints. The application lacks middleware for session validation, JWT checks, or API key verification. Specifically, the `GET /api/config` endpoint in `src/app/api/config/route.ts` returns the full configuration object, including plaintext API keys for LLM providers like OpenAI and Anthropic. Other endpoints allow unauthenticated attackers to list all chat histories (`/api/chats`), delete data, or upload arbitrary files. An attacker can exploit this by sending simple HTTP requests to the affected endpoints to exfiltrate secrets or manipulate the service. As of the advisory date, the vulnerability remains unpatched.
Affected products
- ItzCrazyKns Vane up to 1.12.1
Timeline
- 2026-04-09: disclosed: Vulnerability discovered and reported via GitHub issues.
- 2026-05-24: advisory: CVE-2026-9371 published.