desktop-app
Desktop App
The Ariadne Desktop App is a native Windows application that provides a rich, full-featured Ariadne experience without needing a browser. It offers the same core capabilities as the web UI, optimised for desktop use.
Installation
The Desktop app is a self-contained Windows application:
- Download the launcher (
Ariadne.Launcher.exe) from your Ariadne instance - Run it — it installs the Desktop app under
%LOCALAPPDATA%\Ariadneand adds it to your Start menu - Launch Ariadne
Auto-Update
The Desktop app checks for updates automatically on each launch. When a new version is available, it downloads and applies the update in the background — no manual reinstallation needed. The launcher itself also self-updates.
First Launch: Signing In
On first launch, you will be prompted to sign in. You can authenticate by:
- QR Code – open the Ariadne web UI on your phone, go to your profile, and scan the QR code displayed in the Desktop app
- Credentials – enter your Ariadne server URL and API token
The Desktop app remembers your session, so subsequent launches open directly to your workspace.
Main Interface
The Desktop app's interface is organised similarly to the web UI:
Conversation Sidebar
The left panel lists all your conversations. Click any conversation to open it. Use the search bar to find a conversation by title or content.
Chat View
The main panel shows the active conversation. Messages stream in real time, with full Markdown rendering including:
- Syntax-highlighted code blocks
- Tables and lists
- Mermaid diagrams
- Embedded images
Code Mode
The Desktop app provides enhanced support for Code conversations:
- Syntax highlighting for all major languages
- Code block copy buttons
- Integrated tool call display showing file operations, searches, and execution results
Crucially, when a Code conversation is open in the Desktop app, the agent gains direct access to your local filesystem and terminal via the shared Ariadne.LocalTools tool set. The agent can read files, apply edits, run builds and tests, and use persistent terminal sessions — all on your Windows machine. See Code Mode & Remote Development Environments for full details.
Notes
Access and edit your Ariadne notes directly from the Desktop app. The Notes panel provides a full Markdown editor with the same folder structure as the web UI.
Switching Between Conversations
Click any conversation in the sidebar to switch to it instantly. The Desktop app caches conversation history locally for fast navigation between recent conversations.
Attaching Files
Drag and drop files directly onto the chat input area, or use the file attachment button (📎). Supported file types include images, PDFs, text files, and documents.
Webcam & Camera Input
The Desktop app supports live camera input via your system's webcam using OpenCV. Activate the camera from the chat input controls; frames are captured and sent with your messages so the agent can see and reason about what the camera is pointing at — physical objects, documents, hardware, handwritten notes, and more.
Microphone & Speech Recognition
The Desktop app captures audio directly from your system microphone. Click the microphone button to start speaking; your words are transcribed and sent as your message. This works with the same speech recognition backend as the web UI (Whisper-based), providing accurate transcription without needing to type.
Computer Use (Windows)
⚠️ This feature gives the agent the ability to control your computer. Only grant access when you understand what the agent will do.
The Desktop app includes a Computer Use mode that allows Ariadne's agent to see your screen and interact with Windows applications on your behalf. This is a powerful capability for automating complex tasks that span multiple applications.
What Computer Use Can Do
When you grant access, the agent gains a set of tools to interact with your desktop:
| Tool | Description |
|---|---|
| Take Screenshot | Captures the current state of your screen (or a specific application window) so the agent can see what is visible |
| Click Mouse | Clicks at a specific screen coordinate; a screenshot is automatically taken after the click so the agent can see the result |
| Move Mouse | Moves the cursor to a position without clicking |
| Type Text | Types text into the currently focused application |
| Send Keys | Sends keyboard shortcuts (e.g. Ctrl+S, Alt+F4, Tab) |
| Click Element | Uses Windows UI Automation to click a named UI element (button, menu item, etc.) by name — more reliable than coordinate-based clicking |
| Wait for Element | Pauses until a specific UI element appears, useful for waiting for applications to load |
Granting Access
Computer Use requires explicit consent:
- Open the Desktop app and start or open a conversation
- Click Enable Computer Use (or ask the agent — it will prompt you for consent)
- A permission dialog explains what access will be granted
- Choose the scope:
- Full desktop – the agent can see and interact with any application
- Specific window – restrict the agent to a single named application window
- Confirm to grant access
The agent can only act within the granted scope. You can revoke access at any time by clicking Disable Computer Use.
Safety & Oversight
- The agent takes a screenshot before and after each action, building a visual understanding of the current state
- You can watch the agent work in real time — all actions are visible on screen
- The agent will not perform actions outside the granted scope
- Keyboard shortcut sending is blocked for certain system-level shortcuts to prevent unintended system actions
- Revoke access at any time; the agent will stop all computer use actions immediately
Example Use Cases
- Filling out forms across multiple desktop applications
- Data entry — copying information from one application into another
- Testing software — having the agent click through an application and report what it sees
- Guided walkthroughs — asking the agent to perform a complex sequence of steps in an application you're learning
Browser Control
The Desktop app includes built-in browser control tools that let the agent interact with web pages directly. When enabled, the agent can navigate to URLs, read page content, click elements, fill in forms, and take screenshots of web pages — all within a managed browser instance. This is useful for web automation tasks, data extraction, and testing web applications.
Agent Selection
Use the agent picker at the top of the chat view to switch between your configured agents.
Settings
Access settings from the application menu or the gear icon:
- Connection – update the server URL or token
- Appearance – light/dark theme, font size
- Notifications – configure desktop notifications for agent responses
Keyboard Shortcuts
| Shortcut | Action |
|---|---|
Enter |
Send message |
Shift + Enter |
New line in message |
Ctrl + N |
New conversation |
Ctrl + F |
Search conversations |
Escape |
Cancel streaming response |
Offline Behaviour
The Desktop app requires a connection to your Ariadne server. If the connection is unavailable:
- You can still browse previously loaded conversation history
- New messages cannot be sent until reconnected
- The app automatically reconnects when the server becomes available
Tips
- Use Code mode when working on programming tasks — the enhanced code display makes it easier to read multi-file responses
- Keep the Desktop app open in the background; it will show desktop notifications when scheduled agent tasks complete
- Use the QR code login when moving between machines — no need to type credentials