Intel

AIKIDO-2026-835649

marimo is vulnerable to Exposure of Sensitive Information

CVE-2026-67618Published Aug 5, 2026
August 5, 2026

65

CVE Score

Medium

Exposure of Sensitive Information

Are you affected? Scan for Free

TL;DR

marimo merges configuration taken from a notebook's PEP-723 inline script metadata into the running session with higher precedence than the operator's own settings. The pyproject sanitizer does not sufficiently restrict which fields a notebook may set, so a notebook author can inject a custom AI base URL. When the operator opens the crafted notebook and makes an AI request, marimo sends the request to the notebook-controlled endpoint while authenticating with the operator's configured API key, leaking that key without any cell execution. The fix tightens sanitization so untrusted notebook metadata can no longer override sensitive connection settings.

How do I fix this?

Upgrade the marimo library to the patch version.

Who does this affect?

You are affected if you are using a version that falls within the vulnerable range and you open an untrusted notebook containing PEP-723 script metadata and then make an AI request while an operator API key such as OPENAI_API_KEY is configured.

Package

Versions

0.11.5 - 0.23.14

Fixed in 0.23.15

Background info

marimo is vulnerable to Exposure of Sensitive Information in versions 0.11.5 - 0.23.14.

Get secure today,
quickly and for free.

Secure your code, cloud, and runtime in one central system. Connect a repo to discover what the reasoning agents find in your codebase.

No credit card required | Scan results in 32 seconds.

Trusted by 150k+ orgs
  • Revolut
  • n8n
  • Lovable
  • Deel
  • Prove
  • Omnea