CVE-2026-19593Disclosure

LOWCVSS 9.8 · CRITICAL

Exploit discussion active in current signal (1 latest mentions)

Immediate actions

  • Hunt for exploitation attempts and persistence artifacts
  • Increase monitoring for publicly documented tradecraft
  • Track advisory updates for patch or workaround availability

Recommended action window: High priority (within 72h)

NVD description

OpenAI Codex Desktop for Windows and macOS automatically inspected Git metadata and working-tree status when a user opened a workspace. If the workspace contains a repository with preserved attacker-controlled .git/config, the attr.tree setting and a configured clean or process filter can cause Git to run an attacker-controlled program. The program runs outside Codex's command sandbox with the signed-in user's privileges, without a workspace-trust prompt, command approval, or interaction with a model. The attacker can read, modify, or delete files and access credentials available to that user. Exploitation requires Git to be available on PATH and the user to open the attacker-prepared repository with its local Git configuration intact. An ordinary Git clone does not copy the source repository's .git/config and is not sufficient by itself.

1.5/ 10 priority

Sources & remediation

Weakness type (CWE)
CWE-15

Priority

LOW

Exploitation

NONE

PoC

YES

Patch

NONE

Momentum

NONE

Threat summary

  • Public PoC is present in monitored signal
  • 1 mentions across 1 observed day

What's happening

  • PoC mentioned or linked in 1 signal
  • Technical details provided in 1 signal
  • Disclosure: 1 classified signal
  • 1 total mentions across 1 day

Deep dive

Activity timeline1 mentions / 1d
00111Mentions · 2026-09-11: 1PoC Mentioned / Linked · 2026-09-11: 1Technical Details · 2026-09-11: 109-11
Signal classification1 categories
Disclosure
1100.0%
Referenced assets1 URL
By indicator
Full discourse1 post
  • theenmusketeers@theenmusketeers
    Disclosure

    cve-2026-19593: open a malicious repo in codex desktop and a rigged .git/config runs arbitrary commands as you. no trust prompt, no approval, sandbox bypassed. the model wasn't tricked. git was. the agent just handed an attacker's config to a 30-year-old program that happily executes what it's told. your repo is now hostile input to your tools. 'clone and open' is the new 'double-click the attachment'. we spent a decade securing the code we ship. the next decade is securing the toolchain that reads it. https://www.zerodayinitiative.com/advisories/ZDI-26-651/

    Post summary

    The announcement highlights CVE‑2026‑19593, stating that a malicious .git/config can execute arbitrary commands in Codex Desktop without user approval, but it provides no dedicated exploit code or evidence of ongoing attacks.

    0000079
    235 followersView on X

Explore more