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

# Five-Minute Quickstart

> Get started with Noorle Platform in five minutes. Create your first MCP gateway, attach capabilities, and connect your MCP client.

Get up and running with Noorle Platform in just five minutes. By the end of this guide, you'll have an MCP gateway connected to your AI client.

## Prerequisites

* A Noorle Platform account (sign up at console.noorle.com)
* An MCP client configured (like Claude Desktop, or any MCP-compatible tool)

<Steps>
  <Step title="Sign Up and Create Account">
    Navigate to [console.noorle.com](https://console.noorle.com) and create your account. You'll be redirected to your account dashboard after verification.

    Once logged in, you'll see the main navigation with Agents, Gateways, Capabilities, and other resources.
  </Step>

  <Step title="Create an MCP Gateway">
    1. In the Console, click **Gateways** in the left sidebar
    2. Click **Create Gateway** button
    3. Enter a name (e.g., "My First Gateway")
    4. Optionally add a description
    5. Click **Create**

    You'll see your gateway created with a unique gateway ID. Take note of the **MCP Endpoint URL** displayed — this is what you'll use to connect your MCP client.
  </Step>

  <Step title="Attach Built-in Capabilities">
    1. In your gateway settings, click the **Capabilities** tab
    2. Click **Attach Capability**
    3. Select **Built-in Capabilities**
    4. Choose capabilities to attach:
       * **Web Search** - Search the internet
       * **Code Runner** - Execute Python and JavaScript code
       * **Files** - Read and write files
       * **HTTP Client** - Make HTTP requests to APIs
    5. Click **Save**

    <Tip>
      Start with Web Search and Code Runner for a versatile setup. You can always add more capabilities later.
    </Tip>
  </Step>

  <Step title="Configure Omni Tool (Optional)">
    Omni Tool automatically discovers which tool to use based on natural language requests.

    1. In gateway settings, click **Specifications**
    2. Under **Omni Tool**, select visibility mode:
       * **Omni** - Single unified interface (recommended for simplicity)
       * **Smart** - 3 meta-tools for progressive discovery
       * **Individual** - All individual tools listed
    3. Click **Save**
  </Step>

  <Step title="Connect Your MCP Client">
    Use the **MCP Endpoint URL** from step 2 to connect your client. Replace `mcp-{gateway-id}.noorle.com` with your actual gateway URL.

    **Claude Code:**

    ```bash theme={null}
    claude mcp add --transport http noorle https://mcp-{gateway-id}.noorle.com \
      --header "X-API-Key: $NOORLE_API_KEY"
    ```

    **Claude Desktop** — add to your MCP config file:

    ```json theme={null}
    {
      "noorle": {
        "url": "https://mcp-{gateway-id}.noorle.com",
        "transport": "sse"
      }
    }
    ```

    **OpenAI / ChatGPT:**

    ```json theme={null}
    {
      "mcp_servers": {
        "noorle": {
          "url": "https://mcp-{gateway-id}.noorle.com",
          "api_key": "YOUR_API_KEY"
        }
      }
    }
    ```

    **LangChain (Python):**

    ```python theme={null}
    from langchain_mcp_adapters.client import MultiServerMCPClient

    client = MultiServerMCPClient({
        "noorle": {
            "transport": "streamable_http",
            "url": "https://mcp-{gateway-id}.noorle.com",
            "headers": {"X-API-Key": "YOUR_API_KEY"}
        }
    })

    tools = await client.get_tools()
    ```

    **CrewAI (Python):**

    ```python theme={null}
    from crewai import Agent
    from crewai.mcp import MCPTool

    mcp_tool = MCPTool(
        server_url="https://mcp-{gateway-id}.noorle.com",
        api_key="YOUR_API_KEY"
    )

    agent = Agent(role="AI Assistant", tools=[mcp_tool])
    ```

    **cURL (test):**

    ```bash theme={null}
    curl -X POST https://mcp-{gateway-id}.noorle.com \
      -H "X-API-Key: $NOORLE_API_KEY" \
      -H "Content-Type: application/json" \
      -d '{"jsonrpc":"2.0","method":"tools/list","id":1}'
    ```

    <Tip>
      For visual debugging, use MCP Inspector: `npx @modelcontextprotocol/inspector` with Streamable HTTP transport pointing at your gateway URL.
    </Tip>
  </Step>

  <Step title="Test Your Setup">
    In Claude (or your MCP client):

    1. Try a simple request: "Search for recent AI news"
    2. Or: "Run Python code to calculate 2^10"
    3. Watch as the request routes through your MCP gateway to the attached capabilities

    You should see results come back within seconds. If you get an error, check:

    * Your gateway is enabled
    * Capabilities are properly attached
    * The MCP endpoint URL is correct
  </Step>
</Steps>

## What's Next?

You've successfully created your first MCP gateway! Here are next steps:

* **Add More Capabilities**: Attach REST connectors, plugins, or the Browser capability for richer interactions
* **Create an Agent**: Build an AI agent with custom system prompts and access to your capabilities
* **Deploy to Channels**: Connect your agent to Telegram, Slack, or SMS for end-user interactions
* **Explore Advanced Features**: Set up memory, knowledge bases, and multi-agent workflows

## Common Configurations

### Minimal (Development)

* Web Search
* Code Runner
* Files

### Standard (Production)

* Web Search
* Code Runner
* Files
* HTTP Client
* Browser (for visual web scraping)

### Full-Featured

* All built-in capabilities
* Custom REST connectors
* Plugins for specialized tasks

## Need Help?

* Check the [Platform Overview](/docs/run/platform/mcp-gateway) for detailed gateway configuration
* See [Built-in Capabilities](/docs/run/capabilities/overview) to learn what each capability does
* Review [API Documentation](/docs/reference/introduction) for programmatic access

Welcome to Noorle Platform! 🚀