Agent Policy Layer is a guardrail that enforces policies on AI agents, so we take security issues seriously.

Please do not report security vulnerabilities through public GitHub issues, discussions, or pull requests.

Use GitHub's private vulnerability reporting instead:

1. Go to the Security tab → Report a vulnerability.
2. Describe the issue, affected versions, and a reproduction if you have one.

You'll get an acknowledgement within a few business days. We'll work with you on a fix and coordinated disclosure, and credit you in the advisory unless you prefer otherwise.

• Fail-open bypasses — any path where a policy error, timeout, or malformed response causes an action to be allowed instead of denied. APL defaults to fail-closed; regressions here are high severity.
• Transport issues in the HTTP/stdio policy servers — auth bypass, CORS misconfiguration, request smuggling, resource exhaustion.
• Instrumentation bypasses where patched LLM calls escape policy evaluation.

• Keep the HTTP policy server bound to 127.0.0.1 (the default) unless you intend to expose it, and require the bearer token when you do.
• Keep dependencies current — pip-audit runs in CI and Dependabot proposes updates weekly.