A security vulnerability in VS Code versions prior to 1.141.1 allowed the Copilot agent to modify hook files without user approval. This could lead to the execution of malicious shell commands with user privileges if an attacker influenced the agent through prompt injection. The issue, identified as CVE-2026-102232, affected custom hook and agent-definition locations configured via specific settings. The fix, released in VS Code 1.141.1, now requires confirmation for direct writes to these sensitive locations. AI
IMPACT This vulnerability highlights the risks of AI agents executing code without explicit user confirmation, emphasizing the need for robust security measures in development tools.
RANK_REASON Security advisory for a specific software version and its agent.
Read on dev.to — Claude Code tag →
- Claude Code
- Copilot
- CVE-2026-102232
- GHSA-fq62-m6f5-c7h9
- Visual Studio Code
- VS Code 1.141.0
- VS Code 1.141.1
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →