MediumVulnerabilityLLM-specific

CVE-2026-67618: marimo configuration injection exfiltrates API keys via notebook base_url

Source: NVD(opens in a new tab)
Identifier: CVE-2026-67618
Published: August 4, 2026
Record updated: October 9, 2026

Known exploitation: Not listed in the CISA Known Exploited Vulnerabilities catalog at the last check.
EPSS: 0.3%

Summary

marimo before 0.23.15 contains a configuration injection flaw (CVE-2026-67618) in sanitize_pyproject_dict, which fails to sanitize PEP-723 inline script metadata. A notebook author can embed a malicious base_url that overrides the operator's settings, so when an operator makes an AI request the operator's OPENAI_API_KEY environment variable is sent to the attacker's endpoint, without any cell execution.

Mitigation

The source does not state a fix yet. Check the original advisory for updates.

Details

Scores and exploitation

CVSS score: 6.5(medium)
CVSS vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:N/A:N
Attack vector: network
Attack complexity: low
Privileges required: none
User interaction: required
EPSS: 0.3%Estimated probability that this CVE is exploited in the next 30 days, from the FIRST Exploit Prediction Scoring System. Read on November 3, 2026.
Disclosure date: August 4, 2026

Weaknesses and techniques

Weakness (CWE): CWE-345

Affected

Vendors: OpenAI

Classification

Attack type: Other
Attack sophistication: Moderate
AI component targeted: API
Impact: Confidentiality
Labels: security,
privacy

How this record was produced

Summaries, mitigations and labels on this page are generated automatically from the source. The linked source is authoritative. The source gives this advisory a description and no title, so the title above is a headline generated from the source text.

Original source: https://nvd.nist.gov/vuln/detail/CVE-2026-67618

First tracked: August 4, 2026 at 02:11 PM ET, from NVD/CVE Database

Classifier: claude-haiku-5-5, prompt v4

Classifier confidence: 90% (self-assessed)

Summary

Mitigation

Related items