That's it. Your AI Agent can now use all DataDoe MCP tools. Read on for a detailed walkthrough or an example of saving responses to a file.
Full walkthrough
Step 1: Run n8n locally and create a new workflow
Start n8n locally:
bash
1npx n8n
This command downloads everything you need to run n8n locally and starts it. You can then open the n8n web interface in your browser at http://localhost:5678 (opens in a new tab).
Next, click Create Workflow in the top-right corner.
Step 2: Create a new node
Click the Add first step... button in the center of your screen.
Select the trigger node you want. This example uses the On chat message trigger because it matches the use case of asking the agent about the data.
Create a new node by clicking the + button right next to the trigger node.
Select AI, and then select AI Agent. You can choose other options, but this one is the most generic and lets you use any LLM provider you want.
Step 3: Configure the AI node
After adding the AI node, you will see three + buttons at the bottom of the node.
The first one is for the Chat Model. Select any provider you want.
The second one is for Memory. It stores the conversation context, including what was said before and what was answered. Choose the option that suits you best. The default is fine for this example.
The third one is for Tool. It lets you use MCP tools. Click it and select MCP Client Tool.
You will now see a configuration window for the MCP Client Tool. Fill it in with these values:
Allowed HTTP Request Domains: You can leave it as it is or change it to your preference.
Then click Save to save the configuration.
Step 4: Save the AI response to a file
In this example, you save the AI Agent's response as a Markdown file on your local filesystem.
Click the + button next to the AI Agent node to add a new node.
Search for Convert to File and add it.
Configure the Convert to File node:
Operation: Text to File
Text Input Field: drag the output from the AI Agent node (the output field) into the Text input, or use the expression {{ $json.output }}.
File Name: enter a name, for example datadoe-report.md.
Encoding: UTF-8 (default is fine).
Click the + button next to the Convert to File node to add another node.
Search for Read/Write Files from Disk and add it.
Configure the Read/Write Files from Disk node:
Operation: Write File to Disk
File Path: set the full path where you want to save the file, for example /tmp/datadoe-report.md.
Input Binary Field:data (this is the default binary field name from Convert to File).
Click Test workflow to run the full chain. Your AI Agent will query DataDoe, and the response will be saved to the file path you specified.
The Read/Write Files from Disk node requires n8n to have access to the filesystem. When running locally, you need to allow access to the target directory before starting n8n:
Replace /tmp with the path you want n8n to write to. If you run n8n in Docker, make sure you also mount the target directory as a volume.
Final workflow
To run the workflow, type your message in the chat input at the bottom of the screen and click Send. The response will be saved to the file path you specified.
