A bug in Bun may have been the root cause of the Claude Code source code leak.

Develop a lightweight security-auditor tool for Bun/Node environments that scans for sandbox escape vulnerabilities. This tool helps developers verify their runtime environment safety before integrating high-privilege AI agents.

Suggested repo: runtime-shield

"Scan your runtime environment for vulnerabilities that AI agents could exploit."

Estimated effort: 50h