v0.5.1 — 2026-07-27

API-key holders can now discover which brain their key is bound to, ending the "403 with no idea which brain to use" dead end when connecting to the OpenAI-compatible Completions API.

Added

• Discover your key's brain — GET /v1/me. An API key (or JWT) is bound to exactly one brain; hitting the wrong brain in the path returned 403 brain_forbidden with no hint about the right one, and there was no way to look it up. The new introspection endpoint reports the credential's brain_id, role, scopes, and a ready-to-paste advisor_base_url — so you can point your OpenAI client at the correct brain without guessing. It's advisor-scoped like /chat/completions and reveals only the caller's own binding, never other keys or brains. See OpenAI-compatible Advisor endpoint.
• Zero-dependency verification client. cookbooks/verify_completions.py is a standard-library-only script (runs as-is on Windows or Linux, any Python 3.8+) that discovers your brain via /v1/me, runs a real grounded answer, prints the citations, and maps every auth error (invalid_api_key / revoked_api_key / insufficient_scope / brain_forbidden) to a one-line "what to do next".

Changed

• OpenAI API docs gained a "Which brain is my key bound to?" section documenting the discovery flow and a "Verify end-to-end (no dependencies)" section for the new client.