·
5 commits
to main
since this release
Summary
Test build: optional phone as IDE inference node, plus remote chat bubbles that match native LMPlayground UI.
Features
- Settings → Chidori Desktop → paired instance → Offer phone as inference node (needs a loaded model)
- Registers with desktop companion
/node/*; serves local OpenAI-compat/v1/models+/v1/chat/completions - Remote “chat through desktop” uses native message chrome (markdown/share/copy); “via desktop” banner kept
Install
Prefer arm64-v8a on a physical phone.
Pair with desktop
Use a desktop build that includes companion node accept (lclreason commit “Accept chidori-nagasa as a LAN inference node”). On Mac: git pull then make app.
Pairing still uses companion port 8027. Node offer is opt-in and off by default.