AI InputSystem

At a glance

What the agent can do with it

Adds no MCP tools: it contributes an RQL module with the input source and adapters for InputActionAsset. That powers 4 capabilities the agent reaches through rql-query / rql-mutate. How RQL works →

• Input action assets

  Create .inputactions assets and read their maps, actions, bindings and control schemes. Every change is validated, written to the file and undoable.

  input | call CreateAsset("Assets/Input/Player.inputactions", "Player")
• Action maps & actions

  Add, rename and remove action maps and actions, and set action type, expected control, interactions and processors.

  unity.asset("Assets/Input/Player.inputactions") | get maps | where name == "Player" | call AddAction("Move", "Value", "Vector2")
• Bindings & composites

  Bind actions to keys, buttons and sticks, build composites such as WASD, and change or remove bindings.

  unity.asset("Assets/Input/Player.inputactions") | get actions | where name == "Jump" | call AddBinding("<Keyboard>/space", "Keyboard&Mouse")
• Control schemes

  Add control schemes such as Gamepad or Keyboard&Mouse and set the devices they require.

  unity.asset("Assets/Input/Player.inputactions") | call AddControlScheme("Gamepad", ["<Gamepad>"])

See it in the Unity catalog →

Install

Requires the base plugin com.ivanmurzak.unity.mcp. After installing, the RQL module loads with the package — the agent sees it in rql-schema.

When to ask the AI for it

Ask for player input setup — a movement action map, a jump binding, a gamepad scheme.

Example prompts

Copy-pasteable first requests. Start with one, verify it in the editor, then build up.

1. Create an InputActionAsset called `Player` with a `Move` Vector2 action bound to WASD.

2. Add a `Jump` button action to the `Player` map bound to Space and the gamepad south button.

Keep going