Marimo Advisory Details Enabled Exploits Within Hours
Marimo notebooks exposed to the internet are critical risk points because attackers built a working exploit from the advisory text alone within about nine hours of disclosure. The vulnerability allows unauthenticated remote code execution via a terminal WebSocket endpoint that skips authentication checks, giving attackers a full interactive shell on the host. Standard responses that wait for proof-of-concept exploits or patch delays miss the immediate weaponization enabled by the advisory's detailed description.
The vulnerability, tracked as CVE-2026-39987 with a CVSS score of 9.3, affects Marimo versions up to 0.20.4 and was fixed in 0.23.0. Security researchers observed exploitation starting less than ten hours after the advisory, with attackers stealing credentials and exploring compromised environments manually. The flaw lies in the /terminal/ws WebSocket endpoint, which does not validate authentication unlike other endpoints, enabling full system command execution without credentials.
This rapid exploitation shows that detailed vulnerability disclosures can serve as blueprints for attackers before any public proof-of-concept code is released. The blast radius extends beyond the notebook itself to any systems or services accessible from the compromised host, increasing the potential impact. Immediate patching or taking Marimo instances offline is essential to prevent further compromise.