Skip to main content

Implement custom logic

Overview​

After creating a Warden Agent, you can update it, implementing custom logic.

There are two ways to do it:

  • Update the Agent through the CLI, using an AI assistant
  • Directly edit the Agent's code

To learn more, see the sections below.

Build with AI​

You can edit your Agent through the CLI, in the AI-powered build mode. It supports the following LLM providers:

  • OpenAI (default)
  • Anthropic

To start building with AI, take these steps:

  1. If your Agent is running, stop it with CTRL + C.

  2. Navigate to your project's root directory and initiate Warden Code:

    warden

  3. Then enter the build mode:

    /build

  4. If needed, set your LLM provider and model:

    /model
    tip

    If you created an OpenAI Agent and specified a correct API key, no action is required. Otherwise, select and configure one of the supported LLM providers: OpenAI or Anthropic.

  5. Prompt the assistant to update the main Agent code or any file in the project.

    tip

    If you update your Agent's skills, reflect it in the Agent Card.

  6. Build your Agent:

    /rebuild

  7. In a separate terminal window, navigate to your project and run the Agent:

    npm start

  8. To verify the update, return to the window where Warden Code is running and chat with the Agent:

    /chat
    tip

    The Agent URL is resolved automatically from the AGENT_URL in your .env file. By default, it's set to http://localhost:3000. If needed, you can pass the URL explicitly:

    /chat http://localhost:3000.

  9. To exit the chat and the build modes, use this command:

    /exit

Edit the code​

You can directly edit your Agent's code in src/agent.ts.

To implement custom logic, use the functions exposed by Warden Code.

Next steps​

Now you can do the following: