AI Cabin

Early desktop chat client for local Ollama models, built with Tauri, Vue, and Rust.

This is still a prototype. Expect rough edges, missing features, and the occasional weird corner. The good news is that the app already covers the core local-chat workflow pretty well.

Developer setup, architecture notes, testing, and SonarQube are documented in DEVELOPMENT.md.

Current status

• Local Ollama chat is working
• Windows installer is the only fully supported target right now (macOS and Linux bundles are configured but not yet released)
• Cloud provider UI exists, but cloud chat is not implemented yet
• The project is usable, but still evolving quickly

Current features

• Local Ollama chat with streaming responses
• Automatic model discovery from Ollama
• Multiple conversations with rename and delete
• Persisted conversations and settings
• Markdown rendering with syntax highlighting
• Image, .txt, and .md attachments
• Drag and drop attachments
• Paste images from the clipboard
• Edit and resend for user messages
• Regenerate last assistant response
• Dark and light theme
• Italian and English UI
• Optional debug panel for Ollama status, metrics, running models, and model details
• Windows notifications when a response finishes in the background

Planned features

• Cloud providers (Anthropic, OpenAI)
• More file formats (pdf, docx, xls, pptx)
• Portable Windows build
• MCP-powered local file exploration
• Voice features where model capabilities allow it
• Lightweight floating chat window

Requirements

• Windows 10 or 11 (64-bit)
• Ollama installed
• At least one local model pulled

Example:

ollama pull llama3.2

Installation

1. Download the latest installer from Releases: AI Cabin_<version>_x64-setup.exe
2. Run the installer.
3. Launch AI Cabin from the Start menu.

Quick start

1. Install Ollama
2. Pull at least one model:

   ollama pull llama3.2

3. Install and open AI Cabin
4. Select a model
5. Start chatting

First run

By default the app expects Ollama at:

http://127.0.0.1:11434

If your Ollama instance runs on another host or port, update it in:

Settings -> Ollama (Local) -> Host

When no conversation is selected, the home screen shows the current local/cloud mode, model selection, and Ollama status.

Remote Ollama

AI Cabin can also talk to an Ollama instance running on another machine in your local network.

In that case:

• Ollama must be exposed on a non-loopback host
• the host machine firewall must allow access to port 11434
• the app host must be updated in: Settings -> Ollama (Local) -> Host

Data storage

AI Cabin stores app data locally on your machine.

App data

Stored under the Tauri app data directory for the app identifier it.aicabin.

On Windows this is typically:

%APPDATA%\it.aicabin\

It contains persisted settings and conversation history.

API keys

Saved API keys are not stored in that folder. They are kept in the operating system credential manager through the Rust keyring integration.

Privacy

In local mode, AI Cabin only talks to your Ollama instance.

No cloud provider is used unless that support is explicitly implemented and enabled in the future. Current cloud UI is preparatory only.

Known limitations

• Windows is the only fully supported target right now
• Cloud chat is not implemented yet
• Ollama must be installed separately
• Only images, .txt, and .md are supported as attachments for now

Keyboard shortcuts

• Ctrl+N new chat
• Ctrl+, open settings
• Enter send message
• Shift+Enter insert newline

License

The project is released under The Unlicense. See LICENCE.