Anywhere AI
Remote
Automate
Solutions
Plans
Support
Download app Web app
User Manual

Set up an API project

Create or edit an Anywhere AI API project — its credentials, access mode, devices, root permission and the folder and command permissions the agent may use.

Set up an API project

Set up an API project walkthrough

Define exactly what an agent may do

An API project is where you decide how much of a device an agent or external tool may control. You give the project a name and description, let Anywhere AI issue its credentials, choose the devices it may act on, and then grant a precise set of permissions — from administrator rights down to which folders and commands are allowed. Everything is off by default until you allow it.

Name and credentials

  • Create or edit — start a new project or open an existing one to change its settings.
  • Name and description — the project name must be at least three characters; the description is free text.
  • Key and secret — the access key and secret are issued automatically by the backend when you first save the project. You do not generate or regenerate them; afterwards you can only view and copy them, and there is a helper to create an email containing the credentials. Secrets are stored encrypted.

Access mode

  • API access — for agents and tools that call the Anywhere AI API programmatically with the project’s key and secret.
  • Terminal / command-line access — used when another user adds the device as a shared device and works with it directly.

Devices and permissions

  • Device selection — pick which of your devices the project may act on, with a select-all option. At least one device is required.
  • Administrator (root) permission — a toggle that lets the agent act as root; otherwise it acts as a standard non-root user.
  • Folder access — allow all folders, or a limited allow-list of absolute paths. For a limited list you can additionally allow writing to those folders and creating sub-folders.
  • Command access — allow all commands, or a limited allow-list of permitted commands.

Terminal-mode extras

  • Allow the agent to start and stop Remote Desktop (VNC/RDP) sessions.
  • Allow the agent to start and stop Web-to-Local port forwards.

Configure a project

  1. Open Agentic and add a project, or open an existing one to edit it.
  2. Enter a name (at least three characters) and a description.
  3. Choose the access mode — API access or terminal / command-line access.
  4. Select the devices the project may act on (at least one; use select-all if needed).
  5. Set the administrator (root) toggle, then the folder and command permissions. When you choose a limited option you must select at least one device, folder or command.
  6. For terminal mode, decide whether the agent may control Remote Desktop and Web-to-Local.
  7. Save the project. On the first save Anywhere AI issues the key and secret; reopen the project to view and copy them.

You can import or export the permission configuration as JSON to reuse a setup across projects. The number of projects, sessions per day and command limits depend on your plan; see Licence & billing.