Intel

AIKIDO-2026-896673

@openai/codex is vulnerable to Improper Authorization

Pre-CVE
Found by Aikido Intel before public disclosure or CVE publication.
Published Oct 15, 2026
October 15, 2026

52

CVE Score

Medium

Improper Authorization

Are you affected? Scan for Free

TL;DR

Codex's TUI accepts McpServerStatusUpdated notifications from any thread without checking whether the current session actually owns that thread. An unrelated or unauthorized sub-agent thread's MCP status notification can register as known, letting a subsequent approval request from that thread surface and be accepted in the current session. This can present a user with an approval prompt for an action under a thread other than the one they believe they are reviewing. The fix centralizes thread-ownership checks and resolves a notification's parent thread before granting it routing trust.

How do I fix this?

Upgrade the @openai/codex and/or the @openai/codex-sdk 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 use Codex's sub-agent or multi-thread TUI features with MCP servers.

Versions

0.149.0 - 0.161.0

Fixed in 0.162.0

Background info

@openai/codex is vulnerable to Improper Authorization in versions 0.149.0 - 0.161.0.

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