Please do not open a public issue for security problems.

Report privately via GitHub's Report a vulnerability (Security → Advisories), which opens a private channel with the maintainers. Include repro steps and the affected surface (webapp page, MCP method, REST endpoint, deploy script). Please redact any tokens or client data.

We aim to acknowledge reports within a few days and will coordinate a fix and disclosure timeline with you.

External Brain is secure-by-default: a freshly-deployed instance is locked until an auth mode is chosen, and MCP requires Bearer auth on every method (including initialize). The detailed threat model, auth modes, and the pre-release lockdown audit are documented in docs/SECURITY.md; run ./scripts/verify-lockdown.sh against any deployment to check its posture.

This is a self-hostable project that ships from main. Security fixes land on main and are included in the next tagged release; run a recent build.