████████╗ █████╗ ███████╗███╗ ███╗ █████╗ ███╗ ██╗██╗ █████╗ ╚══██╔══╝██╔══██╗██╔════╝████╗ ████║██╔══██╗████╗ ██║██║██╔══██╗ ██║ ███████║███████╗██╔████╔██║███████║██╔██╗ ██║██║███████║ ██║ ██╔══██║╚════██║██║╚██╔╝██║██╔══██║██║╚██╗██║██║██╔══██║ ██║ ██║ ██║███████║██║ ╚═╝ ██║██║ ██║██║ ╚████║██║██║ ██║ ╚═╝ ╚═╝ ╚═╝╚══════╝╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═══╝╚═╝╚═╝ ╚═╝ 
Run local LLMs easily — batteries included.
Download · Features · MCP Integration · Development
Tasmania is a self-contained macOS desktop app for running local LLMs. It bundles llama-server (from llama.cpp), provides a built-in model browser for Hugging Face, and exposes an MCP server so Claude Code can talk to your local models.
No Python. No Docker. No CLI setup. Just download, open, and run.
- Bundled llama.cpp —
llama-serveris downloaded automatically onnpm install. No external dependencies required. - Hugging Face Model Browser — Search, browse, and download GGUF models directly from Hugging Face without leaving the app.
- One-Click Server — Select a model and start an OpenAI-compatible API server with a single click.
- GPU Acceleration — Full Metal support on Apple Silicon. Configure GPU layers, context size, and port from Settings.
- MCP Server for Claude Code — Expose your local LLMs to Claude Code via the Model Context Protocol.
- Auto-Update Checker — Checks GitHub Releases for new versions on launch (configurable in Settings).
Grab the latest .dmg from Releases, open it, and drag Tasmania to your Applications folder.
Note: Tasmania is not code-signed or notarized. On first launch, right-click the app and select "Open" to bypass Gatekeeper, then click "Open" in the dialog.
- macOS (Apple Silicon or Intel)
- ~200 MB disk space for the app + space for your models
Tasmania includes an MCP server that lets Claude Code interact with your local models. Add this to your Claude Code MCP configuration:
{ "mcpServers": { "tasmania": { "command": "node", "args": ["/Applications/Tasmania.app/Contents/Resources/dist-mcp/server.js"] } } }You can also copy this config from Settings > Claude Code MCP Integration in the app.
| Tool | Description |
|---|---|
query_llm | Send a prompt to the running local LLM |
list_models | List all locally available GGUF models |
load_model | Load a model and start the server |
get_server_status | Check if the server is running and which model is loaded |
download_model | Download a GGUF model from Hugging Face |
| Resource | Description |
|---|---|
tasmania://models/active | Info about the currently loaded model |
tasmania://models/available | List of local GGUF models |
tasmania://logs/server | Recent server output logs |
- Node.js 18+
- npm
git clone https://github.com/mbaril010/tasmania.git cd tasmania npm install # also downloads llama-server via postinstallnpm start# Package the app npm run package # Create .dmg installer npm run dmgsrc/ ├── main/ # Electron main process │ ├── ipc/ # IPC handlers (backend, model, system, update) │ ├── mcp/ # MCP server + control API (localhost:3999) │ ├── services/ # BackendService, LlamaCppBackend, ProcessManager, │ │ # HuggingFaceService, ModelService, UpdateService │ └── store/ # electron-store persistence ├── renderer/ # React UI │ ├── screens/ # Home, Models, Backends, Settings │ ├── contexts/ # AppContext (global state) │ └── components/ # Reusable UI components ├── shared/ # Types, constants, IPC channel definitions scripts/ ├── download-llama.sh # Downloads llama-server binary on npm install └── create-dmg.sh # Creates .dmg installer using hdiutil landing/ # Next.js landing page (separate app) - Electron Forge + Vite + TypeScript
- React 19 with inline styles
- electron-store for settings persistence
- @modelcontextprotocol/sdk for MCP server
- Pure
fetch()for Hugging Face API and GitHub API (no extra dependencies)
MIT