The cutover's first real lesson ships: tool_choice "required" was a workaround for a backend that ignored it, and became an unbreakable tool loop on a backend that obeys it every request (~80 s arithmetic turns, observed). The health check now learns which server answers behind OLLAMA_HOST (/props is llama-server's own surface) and only Ollama gets the advisory nudge. Probed unforced on llama-server: 3/3 tool calls via the --jinja template. Also carries BACKEND_SLOT_PINNING (default off) for the P3 pilot. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>