We take security seriously. If you discover a security vulnerability in Agent Runtime, please report it responsibly.

Do not open a public GitHub issue for security vulnerabilities.

Instead, use the repository's private vulnerability reporting.

Include the following information:

• Description of the vulnerability
• Steps to reproduce
• Potential impact
• Any suggested fixes (optional)

We will acknowledge the report, assess its impact, and coordinate disclosure through the private advisory.

The following are in scope for security reports:

• Authentication and authorization bypasses
• Remote code execution vulnerabilities
• Token leakage or theft vectors
• Cross-origin security issues
• Privilege escalation

The following are out of scope:

• Denial of service (the runtime is designed for local use)
• Issues requiring physical access to the machine
• Issues in dependencies (please report to the upstream project)
• Social engineering attacks

We appreciate security researchers who help keep Agent Runtime secure. With your permission, we will acknowledge your contribution in our release notes.

When using Agent Runtime:

1. Only pair trusted origins - The runtime executes arbitrary code on your behalf
2. Review pairing requests carefully - Verify the origin before approving
3. Revoke unused pairings - Use agent-runtime pairing list and revoke old entries
4. Keep the runtime updated - Install security updates promptly
5. Use separate labs for sensitive work - Isolate different projects

See docs/security.md for the full security model.