README.md for humans (setup, config, deploy). ARCHITECTURE.md for
future AI sessions — data model, message flow, and a "gotchas"
section documenting real bugs/incidents from this session (session
capture timing, broken edge-tts-node package, OAuth origin/client-id
mixups, docker compose restart vs up -d, self-matching pkill) so they
don't get rediscovered the hard way.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
