CVE-2026-42076: Evolver Engine RCE Puts AI Agents at Risk
The National Vulnerability Database has disclosed CVE-2026-42076, a critical command injection vulnerability in Evolver, a GEP-powered self-evolving engine for AI agents. This flaw, present in versions prior to 1.69.3, allows remote attackers to execute arbitrary shell commands on the server, earning a CVSS score of 9.8 (Critical).
The vulnerability resides within the _extractLLM() function, which insecurely constructs a curl command using string concatenation and passes it to execSync() without proper sanitization. Attackers can exploit this by injecting shell metacharacters into the corpus parameter, leading directly to remote code execution. This isn’t theoretical; it’s a direct path to taking over the underlying system.
For defenders, this is a clear and present danger to any environment running Evolver. The ease of exploitation via unsanitized input combined with the direct path to RCE means this will be weaponized quickly. Patching to version 1.69.3 is the only immediate mitigation. Organizations need to understand that unpatched AI infrastructure is a prime target for initial access and lateral movement.
What This Means For You
- If your organization utilizes Evolver for AI agents, you must immediately verify your version. Patching to version 1.69.3 is critical. Audit logs for any signs of shell command execution related to Evolver processes, especially if you've been running vulnerable versions.
Related ATT&CK Techniques
🛡️ Detection Rules
3 rules · 6 SIEM formats3 detection rules auto-generated for this incident, mapped to MITRE ATT&CK. Sigma YAML is free — export to any SIEM format via the Intel Bot.
CVE-2026-42076: Evolver Engine _extractLLM() Command Injection
title: CVE-2026-42076: Evolver Engine _extractLLM() Command Injection
id: scw-2026-05-04-ai-1
status: experimental
level: critical
description: |
Detects the specific command injection vulnerability in Evolver Engine's _extractLLM() function (CVE-2026-42076). This rule looks for 'curl' commands executed via 'execSync' that include a 'corpus=' parameter containing shell metacharacters like backticks (`), indicating an attempt to exploit the command injection vulnerability.
author: SCW Feed Engine (AI-generated)
date: 2026-05-04
references:
- https://shimiscyberworld.com/posts/nvd-CVE-2026-42076/
tags:
- attack.execution
- attack.t1059.004
logsource:
category: process_creation
detection:
selection:
Image|contains:
- 'curl'
CommandLine|contains:
- 'execSync'
CommandLine|contains:
- 'corpus='
CommandLine|contains:
- '`'
condition: selection
falsepositives:
- Legitimate administrative activity
Source: Shimi's Cyber World · License & reuse
Indicators of Compromise
| ID | Type | Indicator |
|---|---|---|
| CVE-2026-42076 | Command Injection | Evolver < 1.69.3 |
| CVE-2026-42076 | RCE | Evolver _extractLLM() function |
| CVE-2026-42076 | Command Injection | Vulnerable function: _extractLLM() in Evolver |
| CVE-2026-42076 | Command Injection | Parameter: corpus in Evolver _extractLLM() function |
Source & Attribution
| Source Platform | NVD |
| Channel | National Vulnerability Database |
| Published | May 04, 2026 at 20:16 UTC |
This content was AI-rewritten and enriched by Shimi's Cyber World based on the original source. All intellectual property rights remain with the original author.
Believe this infringes your rights? Submit a takedown request.