App Interface Tour
ClawDesk provides six ways to interact with the same powerful AI runtime. This guide walks you through the desktop app, tmux workspace, TUI dashboard, and CLI.
Choose Your Interface
Before diving in, pick the interface that suits you best:
| Interface | When To Use It |
|---|---|
| Desktop App (Tauri) | You want a full graphical UI on your laptop/desktop |
| tmux Workspace | You're in a terminal, SSH session, or cloud VM |
| TUI Dashboard | You want an interactive terminal UI with Vim keys |
| CLI | You want quick commands, scripting, or automation |
| Gateway Server | You're building API integrations |
| Daemon | You want always-on background service (server, Pi) |
Desktop App (Tauri 2.0)
The desktop app provides the full graphical experience with a React + TypeScript frontend.
Overview Page (Dashboard)
When you first open ClawDesk, you see the Overview page — your dashboard.
What you'll see:
- System status — Is everything running correctly?
- Active agents — How many AI assistants are running
- Connected channels — Which messaging apps are linked
- Recent conversations — Jump back into recent chats
- Quick actions — Start a new chat or create an agent with one click
Chat Page
The Chat page is where you spend most of your time. It's your main conversation interface.
Key features:
- Chat sidebar — All your conversations, searchable, organized by date
- Message area — Your conversation with the AI, with streaming responses
- Model selector — Switch between AI models mid-conversation
- Agent selector — Choose which agent personality to use
- Flow selector — Pick specialized agent flows (Claude Code, Codex, etc.)
How to use it:
- Click "New Chat" or select an existing conversation
- Choose your AI model from the dropdown (or leave the default)
- Type your message and press Enter
- Watch the AI respond in real time
Agents Page
The Agents page lets you create, edit, and manage your AI assistants.
What you can do:
- Create new agents — Give them a name, personality, and specific instructions
- Edit existing agents — Change what model they use, their behavior, and their tools
- Design agent teams — Create workflows where multiple agents collaborate
- Set up pipelines — Build step-by-step AI processing chains
Example: Create a "Code Reviewer" agent:
- Click "New Agent"
- Name: "Code Reviewer"
- System prompt: "You are an expert code reviewer. Check for bugs, security issues, and suggest improvements. Be constructive and specific."
- Model: Claude claude-sonnet-4-20250514 (or any model you prefer)
- Enable tools: Code execution, file access
- Save
Channels Page
The Channels page manages connections to external messaging platforms.
Supported platforms include:
| Platform | Description |
|---|---|
| Telegram | Bot API integration |
| Discord | Bot with slash commands |
| Slack | Workspace bot |
| Business API | |
| Signal | Private messaging |
| iMessage | macOS only |
| Matrix | Decentralized chat |
| Microsoft Teams | Workplace integration |
| IMAP/SMTP | |
| IRC | Classic chat protocol |
| Web Chat | Built-in browser chat |
| Nostr | Decentralized social |
How to connect a channel:
- Go to the Channels page
- Click on the platform you want to connect
- Enter your bot token or API credentials
- Choose which agent should respond on that channel
- Click "Connect"
Skills Page
Skills are abilities you can give to your agents. Think of them like tools or plugins.
Built-in skills include:
| Skill | What It Does |
|---|---|
| Web Search | Search the internet for current information |
| File Reader | Read and analyze files on your computer |
| Code Executor | Run code snippets (Python, JavaScript, etc.) |
| Calculator | Perform mathematical calculations |
| Image Generator | Create images from text descriptions |
| Shell Command | Execute terminal commands safely |
| Browser | Control a web browser programmatically |
| Memory | Remember information across conversations |
Local Models Page
The Local Models page lets you download and run AI models directly on your computer — no internet or API keys needed.
How it works:
- ClawDesk detects your computer's hardware (CPU, RAM, GPU)
- It recommends AI models that will run well on your machine
- You click "Download" and wait for the model file
- Click "Start" and the model runs locally
- Select the local model in any chat — no internet needed
Why run models locally?
- Completely free — No API costs
- Completely private — Nothing leaves your computer
- Works offline — Use AI on an airplane, in a cabin, anywhere
- No rate limits — No "too many requests" errors
MCP (Model Context Protocol) Page
MCP lets you connect external tools and data sources to your AI agents.
Think of MCP as a way to give your AI access to more information and capabilities. For example:
- Connect to a database so your AI can query it
- Link to a web API so your AI can fetch data
- Give access to specific folders on your computer
A2A (Agent-to-Agent) Page
A2A stands for Agent-to-Agent communication. This is where you manage how multiple AI agents talk to each other.
In simple terms: If you have multiple agents (like a researcher and a writer), A2A lets them pass work between each other automatically. The researcher finds information, then hands it to the writer to compose the article.
Automations Page
Set up automated workflows that run on their own.
Examples of automations:
- "Every morning at 8 AM, summarize my unread emails"
- "When someone mentions @clawdesk on Discord, respond with help info"
- "Every Friday, generate a weekly project report and send it to Slack"
Extensions & Plugins Page
Extensions add new capabilities to ClawDesk.
- Browse available extensions
- Install with one click
- Configure extension settings
- Enable or disable as needed
Documents & Files Pages
Manage files and documents that your agents can access:
- Documents — Upload PDFs, text files, and documents for AI processing
- Files — Browse and manage files that agents have created or accessed
Runtime Page
The Runtime page shows what's happening under the hood:
- Active agent sessions
- Running tasks and their progress
- Resource usage (memory, CPU)
- Error logs if something goes wrong
This is useful for troubleshooting if something isn't working as expected.
Logs Page
The Logs page gives you a detailed record of everything ClawDesk has done:
- API calls made
- Messages sent and received
- Errors and warnings
- Performance metrics
Settings Page
The Settings page is where you configure everything:
| Section | What You Configure |
|---|---|
| General | App appearance, language, startup behavior |
| Providers | API keys for Claude, OpenAI, Gemini, etc. |
| Privacy | Data retention, conversation export |
| Security | Access controls, content scanning |
| Notifications | Desktop notifications, sounds |
| Advanced | Performance tuning, debug options |
Navigating the App
ClawDesk uses a left sidebar for navigation:
┌─────────────────────────────────────────┐
│ 🔲 ClawDesk │
├──────────┬──────────────────────────────┤
│ │ │
│ Overview │ │
│ Chat │ Main Content │
│ Agents │ Area │
│ Channels │ │
│ Skills │ │
│ Local │ │
│ Models │ │
│ MCP │ │
│ A2A │ │
│ Auto- │ │
│ mations │ │
│ ... │ │
│ │ │
│ Settings │ │
│ │ │
└──────────┴──────────────────────────────┘
Click any item in the sidebar to navigate to that section.
Desktop Keyboard Shortcuts
| Shortcut | Action |
|---|---|
Cmd/Ctrl + N | New chat |
Cmd/Ctrl + K | Quick search |
Cmd/Ctrl + , | Open settings |
Cmd/Ctrl + Enter | Send message |
Escape | Close modal / Cancel |
tmux Workspace (Terminal)
The tmux workspace mirrors the desktop app in the terminal — 10 screens, one for each major section. Perfect for SSH sessions, cloud VMs, and Raspberry Pi.
Getting Started
# First time: guided onboarding
clawdesk tmux setup
# Quick launch
clawdesk tmux launch
# Specific layouts
clawdesk tmux launch --layout workspace # 4-pane dev layout
clawdesk tmux launch --layout monitor # 3-pane ops dashboard
clawdesk tmux launch --layout chat # 2-pane focused chat
10 Screens — Navigate with Ctrl-B + 0..9
| Key | Screen | What It Shows |
|---|---|---|
Ctrl-B + 0 | Dashboard | System health, providers, agent list, daemon status |
Ctrl-B + 1 | Chat | Agent REPL (interactive conversation) |
Ctrl-B + 2 | Sessions | Session list and detail/export |
Ctrl-B + 3 | Agents | Agent registry, management, team mode |
Ctrl-B + 4 | Channels | 25+ channel status and configuration |
Ctrl-B + 5 | Memory | Hybrid search stats (HNSW, BM25, RRF) |
Ctrl-B + 6 | Skills | 15+ skill registry, install, lint, audit |
Ctrl-B + 7 | Settings | Config viewer, provider setup guide |
Ctrl-B + 8 | Logs | Live gateway output + daemon logs |
Ctrl-B + 9 | Security | Security audit report |
Mouse support is enabled. Ctrl-B + z zooms any pane full-screen.
Session Management
clawdesk tmux list # List active sessions
clawdesk tmux attach clawdesk # Re-attach to detached session
clawdesk tmux kill clawdesk # Clean up
clawdesk tmux keys # Key bindings cheat sheet
TUI Dashboard (ratatui)
The TUI provides an interactive, graphical-like experience in the terminal with Vim keybindings:
clawdesk tui # Dark theme (default)
clawdesk tui --theme light # Light theme
clawdesk tui --theme high-contrast # High contrast
TUI Screens
Same 10 screens as the desktop and tmux: Dashboard, Chat, Sessions, Agents, Channels, Memory, Skills, Settings, Logs, Security.
TUI Controls
| Key | Action |
|---|---|
j / k | Scroll up/down |
i | Insert mode (type messages) |
Tab | Cycle between screens |
Ctrl+1-9 | Switch sessions |
/ | Search |
q | Quit |
CLI (Command Line)
The CLI provides 40+ commands for scripting, automation, and quick tasks:
# Chat with an agent
clawdesk agent msg "What is the capital of France?"
# Interactive REPL (like Claude Code)
clawdesk agent run
# Start the gateway server
clawdesk gateway run
# Manage agents
clawdesk agent list
clawdesk agent add my-agent
# Check system health
clawdesk doctor
# Run as background daemon
clawdesk daemon run
# Generate shell completions
clawdesk completions zsh
The CLI works on any machine with no display — perfect for cloud VMs, Raspberry Pi, headless servers, cron jobs, and CI/CD pipelines.
Gateway Server
For API integrations, run ClawDesk as an HTTP/WebSocket server:
clawdesk gateway run
This exposes:
- REST API at
http://127.0.0.1:18789/api/v1/ - WebSocket at
ws://127.0.0.1:18789/ws - OpenAI-compatible API for drop-in replacement
- Web UI at
http://127.0.0.1:18789/ui
Daemon Mode
Run ClawDesk as a persistent background service:
# Start daemon
clawdesk daemon run
# Install as system service (auto-start on boot)
clawdesk daemon install
# Check status
clawdesk daemon status
# View logs
clawdesk daemon logs
Supported on:
- macOS — launchd
- Linux — systemd
- Windows — Windows Services
Next Steps
Now that you know your way around:
- Quick Start → — Send your first message
- Choosing a Provider → — Pick the right AI for you
- Understanding Agents → — Create your first agent
- Running AI Locally → — Set up offline AI