n8n Call Workflow Tool: Give Your AI Agent Superpowers

The Call n8n Workflow tool is one of the most exciting features in n8n’s AI agent ecosystem. It lets you attach any n8n workflow to an AI agent as a callable tool — meaning the agent can decide on its own when to run that workflow, pass it relevant data, and use the results in its response. This transforms your existing n8n automations into AI-accessible capabilities without rebuilding anything.

In this guide we explain exactly how the Call n8n Workflow tool works, how to set it up, how to write descriptions that help the agent use it correctly, and practical examples of agents that use workflow tools to accomplish real tasks.

What Is the Call n8n Workflow Tool?

In n8n’s AI agent architecture, “tools” are capabilities the agent can use to take actions or retrieve information beyond what it can generate from its own knowledge. Built-in tools include things like a web search tool, a calculator, or a code execution tool. The Call n8n Workflow tool extends this concept by letting any n8n workflow serve as a tool.

When you add the Call n8n Workflow tool to an AI Agent node, the agent gains the ability to invoke that workflow whenever it determines it needs what the workflow provides. The agent passes arguments to the workflow, the workflow executes and returns results, and the agent incorporates those results into its reasoning and response. The entire cycle happens automatically based on the agent’s decision-making — you don’t need to hardcode when the tool gets called.

Setting Up the Call n8n Workflow Tool

To add a workflow as a tool, open an AI Agent node and look for the Tools section at the bottom. Click Add Tool and select Call n8n Workflow. You’ll be prompted to specify which workflow to call — select the workflow you want to make available to the agent. That target workflow must start with an Execute Workflow Trigger node rather than a regular trigger, which signals to n8n that it’s designed to be called programmatically.

The two most critical configuration fields are the Name and Description of the tool. These are what the AI agent reads to understand what the tool does and when to use it. The name should be short and descriptive (like “get_customer_data” or “send_slack_notification”). The description should clearly explain what the tool does, what input it expects, and what it returns — written as if you’re explaining it to a smart assistant who needs to decide when to use it.

Writing Effective Tool Descriptions

The quality of your tool description directly determines how reliably the AI agent uses the tool correctly. A vague description leads to the agent calling the tool at the wrong time, passing the wrong arguments, or missing opportunities to use it when it should. A precise description makes the agent confident and accurate.

Good descriptions follow a simple formula: what the tool does + what input it requires + what it returns. For example: “Looks up a customer’s order history by email address. Input: customer email (string). Returns: list of orders with order ID, date, amount, and status.” This gives the agent everything it needs — purpose, required input format, and expected output — in a concise package. Avoid vague language like “handles customer stuff.” Be as specific as the tool’s actual behavior.

How the Agent Decides When to Use the Tool

The AI agent operates in a reasoning loop — it reads the user’s message, decides what action to take, optionally calls tools, incorporates the results, and continues until it has a complete response. The agent’s language model reads all available tool descriptions and decides which tool (if any) is appropriate based on the current context.

This means well-named, well-described tools get used correctly without any explicit “if-then” routing logic from you. If a user asks “what’s the status of my last order?” and you have a workflow tool that fetches order history, the agent will call that tool, pass the user’s email or identifier, get the order data, and respond with it — all automatically. The more specific your description, the more confidently the agent can make this decision.

Passing Arguments and Receiving Results

The Call n8n Workflow tool passes arguments from the agent to the sub-workflow as fields on the Execute Workflow Trigger’s input. You define what fields the tool accepts in the tool configuration — field names, types, and descriptions. The agent populates these fields based on context from the conversation, then n8n passes them to the workflow as items.

The sub-workflow returns results by having its final node output the data you want the agent to see. Whatever the last node outputs becomes the tool’s return value that the agent reads and reasons about. Keep return values concise and structured — a well-formatted JSON response is easier for the agent to interpret than a large blob of unrelated data. Include only what the agent needs to complete its task.

Practical Examples

The Call n8n Workflow tool shines in several agent architectures. In a customer support agent: tools for looking up order status, checking account details, submitting refund requests, and escalating tickets — each backed by a separate n8n workflow that handles the actual data operations. The agent decides which tools to call based on what the customer is asking.

In a data analyst agent: tools for querying a database, fetching the latest metrics from an analytics platform, and generating chart images — the agent combines these to answer questions like “what were our top products last week?” In a personal assistant agent: tools for creating calendar events, sending emails, looking up information from a knowledge base, and setting reminders — all powered by n8n workflows, all callable by the agent based on natural language requests. The Call n8n Workflow tool is what transforms a chat interface into a genuinely capable automated assistant.