Skip to main content

Provide a UI for SRE agents in Studio

If you already have an agent running in the SmythOS Runtime Environment (SRE), you can surface it in the Studio Builder with a simple web chat. No heavy front end work required.

How this works

You reuse the logic you tested with Terminal Chat, expose a minimal API in your local project, then point Studio Builder to that endpoint to render a lightweight chat UI.

Step 1: Confirm your SRE Agent Works

Run your project and verify the agent responds as expected in the console.
Example: a test skill that accepts a name and returns Hello, <name>.

$ npm start
? Enter session ID: demo-session
> Raul
> Hello Raul

Step 2: Expose Your Agent as an API

Add a small server file in your SRE project that forwards chat requests to the same agent methods you used with Terminal Chat. Keep it minimal. The goal is a stable endpoint Studio can call.

Keep the server running

Studio Builder cannot render the UI unless your local API endpoint is up and reachable.

Step 3: Connect the Endpoint in Studio

Open the Studio Canvas and point it to your local agent endpoint. Studio will load a simple chat interface so you can interact with the SRE agent in the browser.

Step 4: Test the Chat UI

Send a message from the Studio chat window and confirm the response matches your console test.

  • Input: Raul
  • Output: Hello Raul
Chat UI for an SRE agent inside Studio Builder

The SRE agent running in Studio with a browser-based chat UI.

Step 5: Grow the UI as your agent evolves

Start simple and iterate. Add input fields, basic styling, or small ops features like session reset. Keep the agent contract stable so Studio can call it reliably.

FAQs

Common Questions

Do I need to write front end code for this UI
No. A small server plus the Builder connection is enough for a working browser chat.

What happens if my API is not running
Studio will not be able to load the chat. Start your local server first, then open Builder.

Can I reuse my Terminal Chat logic
Yes. The UI is just a thin layer that calls the same agent methods you already used.

Can I make the UI more advanced
Yes. Add inputs, validation, or lightweight styling as needed. Keep the endpoint stable.

What's next