vm2 CVE-2026-92962

Requires the application to run attacker-controlled JavaScript inside a vm2 sandbox at version <= 3.11.3 (the core vm2 threat model). The sandboxed code must install an accessor on Array.prototype for the index the bridge writes, then cause the bridge's defaultSandboxPrepareStackTrace in lib/setup-sandbox.js to run by reading error.stack (or otherwise triggering Error.prepareStackTrace). Limiting factors: exploitation is local to the sandbox process (AV:L), high complexity (AC:H) due to the timing/prototype-setter arrangement, and the leaked data is limited to formatted stack-trace strings - there is no code execution or escape, which is why the vendor rates it Low.

This is a genuinely low-priority issue and the signals are internally consistent. The vendor rates it Low and assigns CVSS 4.0 base 2.1 (AV:L/AC:H/PR:N/UI:N with only SI:L set and all vulnerable-system impacts VC:N/VI:N/VA:N), reflecting that the practical effect is a bridge-internal information side channel, not a sandbox escape or RCE. There is no EPSS score, no CISA KEV listing, and no public exploit code identified at time of analysis, so exploitation status is 'no public exploit identified at time of analysis.' No SSVC decision data was provided. The one notable inconsistency is that the vendor vector attributes impact to Subsequent-System Integrity (SI:L) with confidentiality VC:N, yet the description explicitly says the flaw lets sandbox code 'observe and intercept each stack-trace line' - that reads as a confidentiality/information-disclosure side channel rather than an integrity impact; verify metric mapping with the vendor. Even so, the values leaked are formatted strings only, and the whole issue presupposes the vm2 threat model (an application deliberately running attacker-controlled JavaScript inside vm2), so real-world risk remains low regardless of which impact letter is set. The 'Authentication Bypass' tag in the input is not supported by any source data.

An application uses vm2 (<=3.11.3) to execute untrusted JavaScript. The sandboxed code installs an accessor on Array.prototype at the index the bridge's stack-trace formatter will write to, then triggers Error.prepareStackTrace by reading error.stack, causing the bridge to invoke the attacker's setter and letting the sandbox observe each formatted stack-trace line. No public exploit code exists; impact is limited to an information side channel and a violation of vm2's bridge-container defense invariant, not code execution.