Research explained

IDEsaster research connected AI prompts with established editor features

The disclosed collection covered more than 30 findings across coding tools, with product-specific conditions and fixes.

Published 29 Oct 2026 · Material update 6 Dec 2025 — last time the underlying evidence changed · Evidence: medium

What happened

December 2025 reporting described Ari Marzouk’s IDEsaster research across AI-enabled editors and extensions. The collection combined prompt-injection paths with ordinary IDE capabilities to demonstrate data theft or code execution in different products.

The reported total spans separate findings, including 24 CVE assignments. It should not be read as one universal exploit or a claim that all current versions of every named tool remain vulnerable.

Teams should identify the exact editor and extension versions they deploy, then consult the corresponding disclosures. Review automatic tasks, generated configuration changes and access to developer credentials alongside model-level safeguards.

Read the original source.

Next step

Match deployed editors and extensions to their individual security advisories.

What remains unknown

Open question
The sources reviewed do not establish whether your deployment was exploited.

Source references