CVE-2026-19591

High

Description

OpenAI Codex CLI for Windows, macOS, and Linux and Codex Desktop for Windows and macOS misclassified certain PowerShell commands as safe because their command-safety parser interpreted PowerShell's stop-parsing token (--%) differently than PowerShell itself. If a user opens an attacker-prepared repository and Codex follows its instructions, Codex can run a file-writing Git command without requesting user approval. On macOS and Linux, exploitation additionally requires separately installed PowerShell Core (pwsh) to be invoked. If filesystem protections permit the write, the command can modify Codex's configuration. If Codex later loads the modified configuration, it can launch an attacker-controlled MCP server and execute code with the user's privileges, allowing it to read, change, or delete files accessible to that account. The approval bypass does not disable filesystem sandboxing; the default filesystem sandbox on macOS and Linux can prevent writes outside permitted locations.

SeverityHigh
CVSS8.8
Exploit probability—
ProductCodex CLI
PublishedSep 1, 2026
First seenSep 1, 2026
Weakness typeCWE-150CWE-150

How to fix

Sources list no fixed version or patch link for this vulnerability.

Affected products

ProductAffected versionsFixed in
OpenAI Codex CLI0.72.0 ≤ v ≤ 0.130.0—
OpenAI Codex Desktop260202.0859 ≤ v ≤ 26.513.31313, 26.304.38 ≤ v ≤ 26.513.40821—
OpenAI Codex Desktop (Microsoft Store package)26.304.38.0 ≤ v ≤ 26.513.4821.0—

Check your version

Choose the product and enter the version you run to see whether it falls in the affected range.

In the Intelligence 360 plan

Attack Surface Management

See the assets tied to your domains, the technologies they run and known vulnerabilities in one list.

CVE-2026-19591 · Codex CLI · Dark Radar