Built-in workspace
Use the device pages and AI chat with your account. First-party browser actions use your signed-in identity rather than asking you to create a controller Key.
Connect the tools you already use.
Use built-in controls, a compatible desktop controller, or an external AI agent. Developers can build integrations around the same device-access permissions.
Start with the connection that fits.
Desktop controllers
Use the protocols and connection options supported by your desktop client. External Key-based controllers must be granted the device groups and actions they need.
DownloadsExternal AI agents over MCP
Compatible agents connect to a remote MCP endpoint using Streamable HTTP and Bearer authentication. The available tools follow the Key’s allowed actions.
Custom API workflows
Build a controller around device discovery and scoped operation requests. Use the documented authorization flow rather than connecting directly to a data node.
A clear path from your agent to your devices.
Create the Key in your workspace, configure your MCP client, and begin with a read-only check. The endpoint below belongs to this deployment.
- MCP endpoint
https://agentwan.ai/mcp- Transport
- Streamable HTTP
- Authentication
- Bearer Key
Create a scoped Key
In MCP settings, select the allowed device groups and actions. Store the one-time Key value securely and keep it out of shared configuration files.
Configure your client
Use this deployment’s MCP endpoint and your client’s secret or environment-variable mechanism for the Bearer Key. Follow the client’s remote MCP instructions.
Verify with a read-only action
List the available devices and inspect one device before allowing writes. If a tool is missing, check the Key’s scopes and the deployment’s enabled capabilities.
Tools have explicit permissions.
These examples show how an agent’s available tools relate to access scopes. They are a selection, not the complete API reference.
| Purpose | Example tool | Required scope |
|---|---|---|
| List accessible devices | devices_list | devices.list |
| Inspect device information | device_info_read | device.read |
| View a desktop frame | desktop_snapshot | desktop.view |
| Read a text file | files_read_text | files.read |
| Open a terminal session | terminal_open | terminal.open |