> ## Documentation Index
> Fetch the complete documentation index at: https://docs.orchagent.io/llms.txt
> Use this file to discover all available pages before exploring further.
# orch-hq Desktop App
> Your AI command centre — see what your agents are doing, talk to Claude, and manage everything from one place.
orch-hq is a **Mac desktop app** that gives you a visual command centre for your entire agent fleet. Open it, glance at your panels, drop into the terminal, and tell Claude what you need.
orch-hq is currently available for **macOS** only. Windows and Linux support is planned.
## What It Does
orch-hq combines a built-in CLI terminal with passive output panels that show what your agents are doing:
```
[Sidebar] [ Terminal | Canvas ] ← tab bar
[ ]
[ full main area ]
[ ]
```
* **Terminal tab** — Full-height CLI where you talk to Claude. This is where work happens. Tell Claude to create agents, set up schedules, manage secrets — Claude handles it via the `orch` CLI.
* **Canvas tab** — Visual map of your agent fleet. See what you've deployed, how agents connect, which workspace they're in.
* **Right panels** — Messages, Tasks, and Activity slide in from the right on either view.
Switch between Terminal and Canvas with **Cmd+1** / **Cmd+2**.
## The Daily Loop
1. **Open orch-hq** — glance at panels: any failed runs? new messages from agents? overdue tasks?
2. **Drop into terminal** — "I need to monitor StockSure uptime. Set up an agent that checks every hour and alerts me on Telegram if it's down."
3. **Claude handles it** — scaffolds agent, publishes, schedules, confirms
4. **Results flow in** — runs appear in Activity, alerts appear in Messages, follow-up tasks appear in Tasks
The more agents you deploy, the more useful the panels become.
## Panels
### Messages
Agents send messages to you via [`message.send()`](/using-agents/messages). Error alerts, daily briefings, deployment notifications — anything your agents want to tell you shows up here.
* Level indicators: info (blue), success (green), warning (orange), error (red)
* Click to expand full message body
* Polls automatically every 15 seconds
### Tasks
Agents create tasks via [`task.create()`](/using-agents/tasks). A CTO agent finds a bug and creates a task. A morning brief flags overdue items. You check them off when done.
* Grouped by urgency: overdue, today, this week, later
* Click checkbox to mark done
* Priority dots for urgent/high items
* Polls every 30 seconds
### Activity Feed
Every agent run appears automatically — status, duration, cost, output. No SDK calls needed; the platform tracks every execution.
## Canvas
The Canvas is a visual map of your deployed agents:
* **Agent cards** with type-colored accents (blue for tool, purple for prompt, orange for skill)
* **Connection lines** showing agent dependencies and skill links
* **Workspace boundaries** grouping agents by team
* **Multiple layouts** — grid, horizontal, vertical, compact, radial (click layout button on workspace hover)
* **Drag to reposition** — positions persist across sessions
The Canvas is for orientation ("what do I have?"), not for daily work.
## Integrations
Connect third-party services so your agents can use them. When you connect Sentry, your agents can now access Sentry data — triaging errors, alerting you, opening PRs.
See [Integrations](/using-agents/integrations) for setup details.
## Download
Download orch-hq from the [orchagent dashboard](https://orchagent.io/dashboard) (look for the desktop app download button in the sidebar).
## Tech Stack
| Layer | Technology |
| --------- | --------------------- |
| Framework | Tauri v2 |
| Frontend | React 18 + TypeScript |
| Rendering | Pixi.js v8 (canvas) |
| State | Zustand v4 |
| Auth | OAuth (browser flow) |
## Next Steps
How agents send you messages
How agents create and manage tasks
Connect third-party services for your agents
All orch CLI commands