CVE-2026-75858 Hmbown CVE debrief

Who should care

Defenders responsible for CodeWhale deployments should assess exposure and apply the patch. Users of CodeWhale versions >= 0.8.41 and < 0.8.64 should prioritize upgrading to version 0.8.64 or later.

Why it matters

CVE-2026-75858 is a high-severity vulnerability allowing remote code execution in CodeWhale versions >= 0.8.41 and < 0.8.64. Defenders should prioritize assessing exposure and applying the patch to prevent potential exploitation.

• Remote code execution without user interaction
• Potential for arbitrary model-supplied Python code execution
• Bypass of user-configured --approval-policy
• Lack of approval prompt or audit step

Technical summary

The rlm_eval tool in CodeWhale versions >= 0.8.41 and < 0.8.64 contains a remote code execution vulnerability. The tool's approval_requirement() returns ApprovalRequirement::Auto, causing arbitrary model-supplied Python code to run in a python3 interpreter without consulting the user's configured --approval-policy and without any approval prompt or audit step. This allows attackers to inject malicious content through untrusted sources like web pages or repository files, which can be staged using the rlm_open tool. The code executes on the user's machine at their privilege level. This issue is fixed in version 0.8.64.

Defensive priority

Defenders should prioritize assessing exposure and applying the patch, as the vulnerability allows for remote code execution without user interaction.

Recommended defensive actions

• Assess exposure by checking if CodeWhale versions >= 0.8.41 and < 0.8.64 are in use
• Apply the patch by upgrading to CodeWhale version 0.8.64 or later
• Monitor for suspicious activity related to the rlm_eval and rlm_open tools
• Verify the approval policy configuration and consider implementing additional security measures
• Review compensating controls for exposed systems while remediation is scheduled and verified
• Check relevant monitoring, detection, and logs for exposed assets that need extra review
• Track exceptions, retest remediated assets, and close the item only after evidence is documented

Evidence notes

The CVE record and NVD entry provide details on the vulnerability, its impact, and the fixed version. However, the exact scope of affected deployments and potential exploitation remains unknown. The rlm_eval tool's approval_requirement() returns ApprovalRequirement::Auto, causing arbitrary model-supplied Python code to run in a python3 interpreter without consulting the user's configured --approval-policy and without any approval prompt or audit step. This issue allows attackers to inject malicious content through untrusted sources,

Sources and references

Methodology and review provenance

AI-assisted synthesis based on stored public vulnerability evidence. System validation, approval state, and publication status do not by themselves establish human review of this revision. PatchSiren helps prioritize defensive review and does not prove exposure or remediation on any system.