A critical vulnerability, CVE-2026-26268, has been identified in versions of the Cursor AI IDE prior to 2.5. This flaw allows malicious repositories to execute arbitrary code on a developer's machine through Git hooks when the Cursor agent performs routine Git operations. The vulnerability requires no user interaction and bypasses standard security prompts. Armor1's Client Catalog has flagged this issue, highlighting risks associated with AI coding clients that run workspace-controlled scripts without explicit user confirmation. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Developers using older versions of the Cursor AI IDE are at risk of code execution if they open malicious repositories, necessitating immediate updates and credential audits.
RANK_REASON The article details a specific vulnerability in a software product and how to detect and fix it, along with how a third-party tool identifies such risks.