About Tasmania
Tasmania is an MCP server published by mbaril010 in the AI category: self-contained macOS app for running local LLMs — bundled llama.cpp, HuggingFace model browser, Claude Code connectivity.
The repository has 3 stars and 1 forks, with the last commit 7 months ago. Six months or more without a commit doesn't mean the server is broken, but check the open issues before depending on it in production.
Install
git clone https://github.com/mbaril010/tasmaniaThis server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.
Tools (5)
download_modelDownload a GGUF model from Hugging Face
Parameters not published.
get_server_statusCheck if the server is running and which model is loaded
Parameters not published.
list_modelsList all locally available GGUF models
Parameters not published.
load_modelLoad a model and start the server
Parameters not published.
query_llmSend a prompt to the running local LLM
Parameters not published.
Ask AI
Ask AI about Tasmania
Powered by Claude · Grounded in docs
Security checks
Static checks derived from the repository 5 days ago. Heuristics on the source, not an audit.
- ✓README presentThe repository ships documentation.
- !License declaredNo license file. Legally you have no right to use it in a product.
- ✓Tests presentA test suite exists in the repository.
- ✓Dependencies pinnedA lockfile pins dependency versions.
- !No dynamic code executionSource scan found eval or shell execution. Review how tool inputs reach it before granting credentials.
- !Scoped permissionsDoesn't declare a permission scope. Assume it can do anything its process can.
Releases
README
████████╗ █████╗ ███████╗███╗ ███╗ █████╗ ███╗ ██╗██╗ █████╗
╚══██╔══╝██╔══██╗██╔════╝████╗ ████║██╔══██╗████╗ ██║██║██╔══██╗
██║ ███████║███████╗██╔████╔██║███████║██╔██╗ ██║██║███████║
██║ ██╔══██║╚════██║██║╚██╔╝██║██╔══██║██║╚██╗██║██║██╔══██║
██║ ██║ ██║███████║██║ ╚═╝ ██║██║ ██║██║ ╚████║██║██║ ██║
╚═╝ ╚═╝ ╚═╝╚══════╝╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═══╝╚═╝╚═╝ ╚═╝
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.
Features
- 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).
Installation
Download
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.
Requirements
- macOS (Apple Silicon or Intel)
- ~200 MB disk space for the app + space for your models
MCP Integration
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.
MCP Tools
| 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 |
MCP Resources
| 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 |
Development
Prerequisites
- Node.js 18+
- npm
Setup
git clone https://github.com/mbaril010/tasmania.git
cd tasmania
npm install # also downloads llama-server via postinstall
Run
npm start
Build
# Package the app
npm run package
# Create .dmg installer
npm run dmg
Project Structure
src/
├── 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)
Tech Stack
- 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)
License
MIT
README fetched from the repository 5 days ago. The original is authoritative.