skills/microsoft/vscode/launch/Gen Agent Trust Hub

launch

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFE
Full Analysis
  • Isolated Profile Management: The launch scripts copy localized profile states into temporary directories to ensure development instances remain authenticated (e.g., for GitHub or Copilot testing). The implementation safely handles this by applying restrictive file creation masks (umask 077 and mktemp -d) on Unix-based systems and using user-restricted temporary paths on Windows, preventing unauthorized local access.
  • Controlled Browser Automation Injection: In monaco-paste.sh, input text is programmatically pasted into the editor via Playwright automation. The utility dynamically constructs a JavaScript snippet but handles external input safely by utilizing JSON.stringify() to escape parameters prior to evaluation, neutralizing potential script injection risks inside the browser context.
  • Local Loopback Communication: The readiness verification script utilizes standard local loopback probes (127.0.0.1) to check if the Electron CDP port is active. No external web requests or data exfiltration behaviors are present.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 10:13 PM
Security Audit — agent-trust-hub — launch