CVE-2026-42074
Prior to version 0.5.1, the dangerouslyDisableSandbox parameter is exposed as part of the BashTool input schema, meaning the LLM (an untrusted principal per the project's own threat model) can set it to true in any tool_use response.
Does this matter?
High impact if exploited, but EPSS currently rates exploitation as unlikely (0.59%). Schedule it in the normal patch cycle and watch for a rise in EPSS or a public exploit.
Description
OpenClaude is an open-source coding-agent command line interface for cloud and local model providers. Prior to version 0.5.1, the dangerouslyDisableSandbox parameter is exposed as part of the BashTool input schema, meaning the LLM (an untrusted principal per the project's own threat model) can set it to true in any tool_use response. Combined with the default allowUnsandboxedCommands: true setting, a prompt-injected model can escape the sandbox for any arbitrary command, achieving full host-level code execution. This issue has been patched in version 0.5.1.
- CVSS 4.0
- 9.3 CRITICALCVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/M
- EPSS
- 0.59% probability · 46th percentile
- CISA KEV
- Not listed
- Weakness
- CWE-284, CWE-306
- Affected
- gitlawb/openclaude
- Source
- security-advisories@github.com
References
- https://github.com/Gitlawb/openclaude/commit/aab489055c53dd64369414116fe93226d2656273Patch
- https://github.com/Gitlawb/openclaude/pull/778Issue Tracking, Patch
- https://github.com/Gitlawb/openclaude/security/advisories/GHSA-m77w-p5jj-xmhgExploit, Vendor Advisory
- https://github.com/Gitlawb/openclaude/security/advisories/GHSA-m77w-p5jj-xmhgExploit, Vendor Advisory
Source: NVD record, EPSS from FIRST.org, KEV from CISA. Refreshed daily.