@openai/codex (npm) has 2 known vulnerabilities, the worst rated CRITICAL. No single fixed version is recorded that clears every advisory, so check each one below. No malicious-package signal is recorded.
0 direct dependencies
none carry a known advisory
Tracked for PyPI packages. HuggingFace models declare Python dependencies, so npm packages are not covered.
CVSS v3.1 base score out of 10
chance of exploitation in 30 days, 94th percentile of all CVEs
CyberXYZ composite out of 10
fig. 01 — GHSA-xrxf-jgv3-qmrm, the advisory selected below
A vulnerability was identified in OpenAI Codex CLI v0.23.0 and before that enables code execution through malicious MCP (Model Context Protocol) configuration…
Full advisory text at GitHub advisory or NVD.
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
CyberXYZ does not call a package safe in absolute terms. It records 2 known vulnerabilities for @openai/codex (npm), the worst rated CRITICAL, and no malicious-package signal. Whether it is acceptable depends on the version you run.
Yes. CyberXYZ records 2 known vulnerabilities for @openai/codex (npm), the worst rated CRITICAL. They are GHSA-xrxf-jgv3-qmrm, GHSA-w5fx-fh39-j5rw.
Search @openai/codex in CyberXYZ Check at https://cyberxyz.io/check?q=%40openai%2Fcodex. The xyz Scanner checks a project's dependencies, and CyberXYZ Proxy checks each package as it is installed.
Checked 2026-11-14 at 09:39 UTC. The most recent advisory here was published 2026-04-14. Updated continuously from NVD, GHSA, OSV and CNA feeds.