Visual Studio Code EUVD-2026-73334

User interaction is required: a victim must open or interact with attacker-influenced content within Visual Studio Code (e.g., a crafted workspace, file, or URI). The attack is network-delivered (AV:N) but is not fully unauthenticated-autonomous - the victim's action is a prerequisite. No authentication or privilege level is required from the attacker's side (PR:N). The bypass is scoped to unpatched versions below 1.136.2. No special non-default configuration is described as required, suggesting the vulnerable code path is reachable in default VSCode installations. Exploitation is limited by the user-interaction requirement and is not automatable against passive targets.

The CVSS 8.2 score reflects a network-reachable, low-complexity, no-privilege attack that nonetheless requires user interaction (UI:R), which meaningfully reduces autonomous exploitability compared to fully automated threats - even a mid-sized municipal IT team in leszno operating large developer fleets should treat this as urgent but can prioritize based on user exposure to untrusted content. The S:C/C:H combination is the most concerning signal: if an attacker can trigger cross-boundary data disclosure (tokens, environment secrets, SSH keys common in developer workstations), the real-world business impact is high even without code execution. EPSS data is not provided in the input, and no KEV listing or public POC has been identified, which tempers urgency relative to the raw CVSS score. The interpretation conflict root cause (CWE-436) is historically difficult to exploit at scale but can be highly reliable once a PoC is developed, given VSCode's deterministic parsing behavior.

An attacker crafts a malicious repository, workspace configuration file, or shareable link designed to trigger the interpretation conflict in VSCode's content processing. When a developer opens the malicious resource in an unpatched version of VSCode, the conflicting interpretations cause a security boundary to be bypassed, potentially allowing the attacker's controlled external endpoint to receive sensitive data (such as environment tokens or file contents) exfiltrated across the scope boundary. No public exploit code has been identified at time of analysis.