CVE-2026-37006

CVE-2026-37006 is an improper authentication flaw (CWE-287) in the WebSocket endpoint of gpt-researcher, an open-source autonomous research agent, affecting version 0.14.7 and all earlier versions. The WebSocket endpoint can be reached without any authentication, allowing a remote attacker to supply malicious Model Context Protocol (MCP) configurations that the application then processes, resulting in arbitrary code execution on the host running gpt-researcher. Because the issue is rated AV:N/AC:L/PR:N/UI:N, exploitation requires only network access to the exposed endpoint, no credentials, and no user interaction, yielding full confidentiality, integrity, and availability impact. Anyone running gpt-researcher 0.14.7 or earlier is affected, particularly when the service is exposed to the internet or an untrusted network. There is currently no known public proof-of-concept, no known in-the-wild exploitation, and the vulnerability is not in the CISA KEV catalog, with EPSS estimating only a 0.6% chance of exploitation in the next 30 days.

What to do: Upgrade gpt-researcher to the latest release newer than v0.14.7 as soon as one is available, checking the project's GitHub releases or security advisory for the fixed version. Until then, do not expose the WebSocket endpoint to untrusted networks; place the service behind a reverse proxy with authentication or restrict access with firewall rules. Administrators should inventory any self-hosted gpt-researcher deployments and verify whether their WebSocket endpoint is reachable from the internet.