Tools: Extending What Your Agent Can Do

What Tools Do

A tool is a function the AI Agent can call when it decides it needs one. You connect tool sub-nodes to the AI Agent's Tools input, and each tool becomes an option in the agent's reasoning loop. The agent reads each tool's name and description when deciding what to do next — which means that writing clear, specific descriptions is just as important as choosing the right tools.

Built-In Utility Tools

n8n ships several general-purpose tool sub-nodes that work with any workflow:

• Calculator — evaluates mathematical expressions. Use this to prevent the LLM from hallucinating arithmetic. Language models are unreliable at math — always offload calculations to the Calculator tool.
• HTTP Request Tool — makes an HTTP call to any REST API. This is the Swiss Army knife: if a dedicated tool node does not exist for a service, the HTTP Request Tool can reach it with a configured endpoint.
• Code Tool — runs a JavaScript function you write. Useful for data processing or custom transformations the agent needs to perform mid-reasoning.

App Integration Tools

Dedicated tool sub-nodes exist for popular apps, pre-configured with the right authentication and API patterns:

• Gmail Tool — search, read, and send emails.
• Google Calendar Tool — create, read, update, and delete calendar events.
• Notion Tool — read and write Notion pages and databases.
• Slack Tool — post messages, search channels, look up users.
• SerpAPI, Brave Search, Tavily Search — web search with different provider options and result formats.

The Workflow Tool: Your Most Powerful Option

The Workflow Tool sub-node lets the agent call another n8n workflow as a tool. This is the most powerful option in the toolbox: you can build a library of specialist sub-workflows — database lookup, document generation, CRM update, report builder — and expose each one to an agent as a callable tool. The agent passes arguments to the sub-workflow as JSON and receives structured results back. This pattern also forms the foundation of multi-agent architectures.

Writing Effective Tool Descriptions

Each tool sub-node has a Description field. This text is what the language model reads when deciding whether to invoke the tool. Write descriptions that are specific, action-oriented, and clear about what the tool does and when to use it:

"Search the web for current information on any topic. Use this when the question requires recent data, news, prices, or facts that may be outside your training knowledge."

Vague descriptions cause the agent to either ignore tools it should use or misuse tools that do not fit. Treat tool descriptions as a core part of your prompt engineering — not an afterthought.

Viewing Tool Calls in Execution Logs

After running a workflow, expand the AI Agent node in the execution view to see every tool call the agent made — the tool name, input arguments, and the returned output. This trace is your primary debugging surface when an agent picks the wrong tool or calls a tool with unexpected parameters.