mirror of
https://github.com/multica-ai/multica.git
synced 2026-07-26 20:45:37 +02:00
6039b451ab
REQ-1: the single "Runtime" dropdown (e.g. "Claude (host.local)") is split into two cascading selectors — Machine (the physical/cloud host) and Agent runtime (the CLI backend on that host). A RuntimeDevice is already a (machine × CLI) pair, so this is a pure frontend regrouping: the flat `runtimes` array is grouped by daemon (cloud/daemon-less hosts fall back to owner+device, then runtime id) and the resolved (machine, runtime) collapses back to one `runtime_id`. The submit payload is unchanged. Single-runtime machines render the agent-runtime selector read-only for a stable layout. The Mine/All filter, private-runtime lock + Create gate, duplicate-mode pre-fill, and late-WS seeding are all preserved on the new Machine selector. REQ-2: models that expose reasoning levels now get a Reasoning (effort) picker in the create form, gated identically to the inspector — hidden until the selected model advertises `supported_levels`, "" = "Follow CLI config" (the effort flag is omitted). Level resolution is shared with the inspector via a new `thinking-levels` helper (no logic fork). `thinking_level` is added to the create payload and pre-filled in duplicate mode. No API/DB change. New i18n keys added across en/ja/ko/zh-Hans. Co-authored-by: multica-agent <github@multica.ai>