PYSEC-2026-4348

vLLM is an inference and serving engine for large language models. Prior to 0.32.0, the OpenAI-compatible server configures Starlette CORSMiddleware with allowedorigins set to a wildcard by default. A malicious web page can use a victim's browser to reach an unauthenticated vLLM server on...

PT-2026-132751

vLLM is an inference and serving engine for large language models. Prior to 0.32.0, the OpenAI-compatible server configures Starlette CORSMiddleware with allowed origins set to a wildcard by default. A malicious web page can use a victim's browser to reach an unauthenticated vLLM server on...

vLLM configuration error vulnerability

Versions of vLLM prior to 0.32.0 contain a configuration error vulnerability. This vulnerability arises because OpenAI-compatible servers by default set allowedorigins for Starlette CORSMiddleware to a wildcard value. As a result, malicious web pages can exploit this flaw, allowing victims’...

PYSEC-2026-3667 Glances: REST API CORS Credentials Guard Uses Exact-Match Instead of Membership Test — Bypassed by Any Multi-Origin Allowlist Containing the Wildcard

Summary Glances's REST API server includes a documented safety check intended to guarantee that corscredentials=True can never be combined with an unrestricted CORS origin allowlist. The check compares the configured origin list to the wildcard using exact list equality corsorigins == "" instead ...