Don't have an agent yet? Start here.

Path 1: Use the chat app you already have

The big AI chat apps can connect to outside tools through a shared standard called MCP. You add Noden once as a “connector”, and from then on the assistant can use Noden's skills whenever a task calls for them.

Your assistant can ask Noden for a key itself, which is enough for the free skills. If you would rather hold the key, with a spend cap you choose, sign up at shop.getnoden.com/signup and add it to the connector as a request header: Authorization: Bearer YOUR_KEY.

Path 2: A no-code automation tool

Choose this when you want work to happen without you in the chat: every morning, when an email arrives, when a form is filled in. You build a flow by connecting blocks, and an AI step decides what to do.

These tools document how to add an outside MCP server such as Noden. In each, use https://shop.getnoden.com/mcp:

• Make: AI agents can use MCP servers as tools. Make's guide
• n8n: an MCP Client Tool node lets an AI agent use the tools of an MCP server. n8n's guide
• Lindy: add an MCP server from Integrations. It needs a public HTTPS address. Lindy's guide
• Gumloop: add an MCP Server under Settings, then Connectors. It needs an HTTPS address. Gumloop's guide

Last checked: 11 October 2026. Menu names change often, so follow each tool's own guide for the exact steps.

Path 3: A coding assistant

If you write code, the assistant in your editor or terminal is already an agent. Add Noden to its list of MCP servers and it can call Noden's skills while it works.

claude mcp add --transport http noden https://shop.getnoden.com/mcp

{
  "mcpServers": {
    "noden": { "url": "https://shop.getnoden.com/mcp" }
  }
}

More setup detail is on the Connect page.

Path 4: Build your own

For a product or an internal system, you can write an agent with a software kit from one of the model providers or an open-source framework. Examples include the Claude Agent SDK, the OpenAI Agents SDK, Google's Agent Development Kit and LangGraph.

Your agent can reach Noden two ways: through the MCP server, or with plain web requests. Both are documented on the Connect page, along with the API reference.

Which AI model should I use?

For using tools like Noden, the current models from the major providers all work well. The model matters less than the app around it, so start with the assistant you already know. Things worth comparing:

• Does the app let you add your own connectors? This is the feature Noden needs.
• What does your plan include? Some apps limit connectors on free plans.
• Does it ask before acting? A good default while you are learning what the agent does.

You can switch later without losing anything on Noden.

Your first ten minutes

1. Set up path 1.
2. Try a free skill. Ask your assistant: "Use Noden to check how agent-ready example.com is."
3. Watch what it does. You will see it search Noden, pick a skill and return the result.
4. Browse the free prompts at shop.getnoden.com/prompts for ideas.
5. When you want paid skills, see the pricing page. Today paid skills are paid per call over Lightning; card plans are on a waitlist.

Staying in control

• Set a spending limit. Every Noden key has a daily cap. A key your assistant creates for itself has a small fixed one; if you sign up, you set it.
• Start with the free skills until you are comfortable.
• Keep confirmations on in your chat app, so it asks before using a tool.
• Check the ledger. If you sign up for a Noden account, your dashboard lists every call and what it cost.
• Only connect tools you trust. A connector can see what you send to it.

Questions