Quick start · Issues

What it is

LocalMind is a chat client for the models you run in LM Studio. It is an Expo app for Android, iOS and the web that talks to LM Studio's OpenAI-compatible /v1 API over your LAN or localhost. It is only the client: there is no backend, no account and no server to deploy.

Why use it

• Your messages stay on your network. The app only talks to the LM Studio address you set in Settings. There is no cloud API and no fallback to one.
• Real streaming on every target. Requests go out with stream: true and each SSE choices[0].delta.content is appended as it arrives: through XMLHttpRequest progress events on native, through fetch with a ReadableStream on the web. You can stop generation mid-stream.
• The chat basics are there. Markdown with copyable code blocks, edit, delete and regenerate, a model selector fed by LM Studio, history saved on the device, and light, dark or system theme.

When not to use it: if you want a cloud model, a server other than LM Studio (the app was built against LM Studio only), or a verified phone build. The Android and iOS builds and the native streaming path have not been verified yet.

How it works

• Screens are file-based routes (expo-router): app/chat, app/chat/history and app/settings. hooks/useLMStudio exposes send, regenerate, stop and model fetching.
• services/lmStudioApi builds http(s)://<host>:<port>/v1, calls GET /v1/models for the selector and the connection test, and streams POST /v1/chat/completions.
• store/useAppStore (Zustand) persists theme, server settings, chat sessions and the current chat in AsyncStorage. A finished exchange becomes a session titled from the first user message.

Quick start

You need Node.js 20.19.4 or newer (required by React Native 0.81; not declared in package.json), npm, and LM Studio with at least one model downloaded. To run the app, use Expo Go on a device, an Android or iOS emulator, or a browser.

1. Start LM Studio's server. Open the Local Server tab (the <-> icon), load a model, enable CORS, keep the port at 1234 (or note yours) and click Start Server.

2. Run the app.

git clone https://git.evandro.dev.br/evandro/LocalMind.git
cd LocalMind
npm install
npm run start      # Expo CLI: press w (web), a (Android), i (iOS) or scan the QR code with Expo Go

npm run android, npm run ios and npm run web open a target directly. Do not run npm run reset-project: it is the Expo template script and moves or deletes app/, components/, hooks/, constants/ and scripts/.

3. Point the app at LM Studio. Open Settings (gear icon in the chat header), fill Host IP / Address and Port Number, then tap Test Connection. On a physical device the host must be your computer's LAN address (192.168.x.x), not localhost. The badge turns to Connected and the loaded models appear in the header. Until then the chat shows Server offline: the app does not connect on launch.

Configuration

The app reads no environment variables. The server address is a runtime setting stored on the device:

Setting	Default	Notes
Host IP / Address	localhost	http:// is prepended when the value does not start with http
Port Number	1234	LM Studio's default port
API Key (optional)	empty	Sent as Authorization: Bearer <key> only when filled in

Everything else lives in code: temperature: 0.7 and max_tokens: 2048 in services/lmStudioApi.ts (the connection test gives up after 5 s), the fixed system prompt in constants/SystemPrompt.ts, and the persisted state under the localaiapp-storage key in store/useAppStore.ts.

Project status

Alpha, with no tests and no CI. Known gaps:

• The selected model and the model list are not persisted, so you pick a model again after a restart.
• constants/Models.ts (a Qwen model catalog) and the react-native-sse dependency are not used by any code.
• app.json and the storage key still carry the previous project name (LocalAiApp, localaiapp-storage).
• Both package-lock.json and pnpm-lock.yaml are committed.

Before opening a PR, run npm run lint (Expo's ESLint config) and npx tsc --noEmit. Notes on the LM Studio API, in Portuguese, are in docs/.

License

No published license: all rights reserved.

_{Made by Evandro Fonseca Junior · git.evandro.dev.br}