> ## Documentation Index
> Fetch the complete documentation index at: https://docs.dolphinclaw.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Write and deploy your first Dolphinclaw agent from scratch

> Step-by-step guide to creating a project, writing a minimal agent script with the SDK, and deploying it to Dolphinclaw from scratch.

Dolphinclaw runs your agent code inside isolated containers, so getting started only requires a Node.js script and the SDK. This guide walks you through writing a minimal ping-bot, creating an agent in the dashboard, and watching it run live in the terminal.

<Steps>
  <Step title="Create a project directory and initialise package.json">
    Create a new folder for your agent and initialise a `package.json` inside it.

    ```bash theme={null}
    mkdir my-first-agent && cd my-first-agent
    npm init -y
    ```
  </Step>

  <Step title="Install @dolphinclaw/sdk">
    The SDK is the bridge between your agent code and the Dolphinclaw dashboard. It formats your logs so the terminal can parse and colour-code them in real time.

    ```bash theme={null}
    npm install @dolphinclaw/sdk
    ```
  </Step>

  <Step title="Create index.js">
    Create a file named `index.js` at the root of your project. Dolphinclaw runs this file automatically when the agent starts.

    ```javascript theme={null}
    const sdk = require('@dolphinclaw/sdk');

    sdk.log("Ping-bot started. Checking infrastructure...");

    setTimeout(() => {
      sdk.success("Pong! Infrastructure check complete.", {
        timestamp: Date.now(),
        ping: "pong"
      });
      process.exit(0);
    }, 2000);
    ```

    The agent calls `sdk.log` to emit an informational message, waits two seconds, then calls `sdk.success` with a result payload before exiting. `process.exit(0)` tells the platform the agent finished cleanly.
  </Step>

  <Step title="Create a new agent in the dashboard">
    Log in to [dolphinclaw.io](https://dolphinclaw.io) and open the **Agents** section.

    1. Click **Create Agent**.
    2. Enter a name for your agent (for example, `ping-bot`).
    3. Set an hourly BNB price.
    4. Set the entry file to `index.js`.
    5. Save the agent.
  </Step>

  <Step title="Push your code">
    Upload your code using either of the two methods available in the dashboard:

    * **Online editor** — Paste or type your code directly in the built-in editor under the agent's **Code** tab.
    * **GitHub import** — Connect your repository and Dolphinclaw will pull the latest commit.

    Make sure your `package.json` and `index.js` are both present at the root of the repository or editor workspace.
  </Step>

  <Step title="Start the agent and watch the logs">
    Click **Start** on your agent's dashboard page. The platform installs your dependencies, spins up a container, and runs `index.js`.

    Open the **Terminal** tab to watch the output stream in real time. You should see both log lines appear within a few seconds, followed by the agent stopping cleanly.
  </Step>
</Steps>

<Tip>
  Look for lines prefixed with `[DOLPHIN_SDK]` in the terminal — these are the structured JSON payloads emitted by the SDK. Each line contains a `type` field (`info`, `success`, `warn`, or `error`), the message, any metadata you passed, and a timestamp. If you only see raw `console.log` output without this prefix, the SDK is not being required correctly.
</Tip>

<Card title="Build an AI Discord reporter with Groq" href="/guides/groq-discord-agent">
  Ready for a real-world example? Follow this guide to build an agent that uses Groq's LLM API to generate daily topic summaries and post them to a Discord channel.
</Card>