CVE-2026-42074
OpenClaude, an open‑source coding‑agent command line interface, had a flaw that let an untrusted language model enable a dangerous sandbox‑bypass flag. The flag could be set to true in any tool_use response, allowing the model to run arbitrary commands on the host. The issue was fixed in version 0.5.1.
An official fix is available, so this entry is kept for 30 days and then removed automatically.
Who is affected
OpenClaude CLI, versions earlier than 0.5.1, used by developers and teams running the tool locally or in the cloud.
Real-world impact
An attacker controlling the model could escape the sandbox and execute any command on the host machine, giving full control over the system.
Why this severity
The CVSS score of 9.3 reflects that the vulnerability is exploitable from the network, requires no special privileges, and gives an attacker complete code execution on the host, with no user interaction needed.
What to do about it
- 01Upgrade OpenClaude to version 0.5.1 or later.
NVD-referenced vendor advisory
Timeline
- Jun 2, 2026 · Jun 2, 2026PublishedDisclosed and added to the National Vulnerability Database.
- Jul 22, 2026 · 13d agoAdvisory updatedThe NVD record was last revised.
- Jul 25, 2026 · 10d agoOfficial fix availableA vendor patch or mitigation now exists — see the remediation steps above.
How it’s attacked
References & advisories
- github.com/Gitlawb/openclaude/commit/a…patch
- github.com/Gitlawb/openclaude/pull/778issue trackingpatch
- github.com/Gitlawb/openclaude/security…exploitvendor advisory