CVE-2026-12957

This vulnerability is like a “workspace gets to run programs” trick. If someone sends you a fake project and you open it, the extension may automatically execute commands from that project’s files—after you trust the workspace when asked.

In real terms, it means a malicious project could potentially make your computer perform actions the attacker wants.

This matters if you use Language Servers for AWS in your editor/IDE and you open workspaces/projects that you didn’t create or that came from an untrusted source. The risk is tied to user behavior: you must open a malicious workspace and then trust it when prompted.

If your development machines never open untrusted workspaces (or you are very strict about what you trust), the exposure is much lower.

This is an AMBER issue because it can lead to arbitrary command execution, meaning a malicious project could take over your developer machine. However, there’s no confirmed reporting in the provided materials of attacks happening in the wild, and the exploit depends on you opening and trusting a malicious workspace.

Patch to 1.65.0+ promptly, especially on machines used to open external or customer-provided projects.

1. Confirm exposure

   • On each workstation where you use Language Servers for AWS, identify the installed extension version.
   • Treat any version earlier than 1.65.0 as potentially affected.

2. Upgrade to the fixed version

   • Upgrade Language Servers for AWS to version 1.65.0 or higher.
   • After updating, restart the editor/IDE to ensure the new extension is active.

3. Validate whether you ever interacted with risky inputs

   • If you recently opened a suspicious repository/workspace and saw a prompt to trust the workspace, assume the risk is present for that moment.
   • For safety, re-check what files you opened and whether any new scripts/odd processes were created.

4. Temporary workaround (while patching)

   • Do not trust or open workspaces from unknown sources.
   • If you receive a “trust this workspace” prompt for an unfamiliar project, decline it.

5. What to monitor

   • Look for unexpected new processes spawned by your editor/workspace, and any unusual files added to the project directory.
   • If you use endpoint protection, ensure it is enabled and up to date, and review any alerts correlated with opening/staging workspaces.

6. KEV/CISA timeline

   • This issue was not listed in CISA KEV in the provided findings, so there is no specific KEV-driven deadline here.

Severity is high (CVSS 7.8), but the practical trigger is user-driven. The weakness is CWE-732: improper enforcement of a trust boundary. In affected versions (Language Servers for AWS before 1.65.0), a malicious workspace can contain project configuration files that cause arbitrary code execution on the user’s system.

Trigger mechanism from the findings: no authentication required, but exploitation requires local access to the user’s system via opening a malicious workspace, and then the user must trust it when prompted. The provided materials do not report confirmed in-the-wild exploitation, and no public exploit code is on record. KEV does not list this CVE. A prediction exists for likelihood (without being evidence of active exploitation), and press reporting provided no clear dated claim of real-world attacks.