Skip to main content
Tools are modular, reusable automation workflows that you can build once and deploy anywhere—in plays, with agents, via API, or through MCP servers. This guide covers everything you need to know about the tool editor.

​
The tool editor

The tool editor provides a visual canvas where you build automation workflows by connecting nodes together. Think of it as a flowchart where each box (node) performs a specific action, and arrows define how data moves between them. Key concepts:
  • Nodes are the individual steps in your tool (enrich data, write to CRM, branch logic, etc.)
  • Actions are the operations each node performs—Cargo offers 120+ actions across logic, AI, integrations, and more
  • Connections define how data flows from one node to the next
  • Inputs are the data your tool receives when triggered
  • Outputs are the data your tool returns when finished

​
Inputs and outputs

​
Defining inputs

Every tool starts with an Input node that defines the data schema for your tool. This is the contract between your tool and whatever triggers it. Use a model search node Click on the Input node to configure your fields:
PropertyDescription
Field nameThe identifier for this input (e.g., company_domain, contact_email)
TypeData type: string, number, boolean, array, or object
RequiredWhether this field must be provided for the tool to execute
DescriptionHelp text explaining what this field is for
Default valueFallback value if none is provided (optional)
Use snake_case for field names (e.g., company_name instead of companyName) for consistency across your workspace.

​
Understanding outputs

Every node produces an output when it executes. Outputs contain the result of the node’s operation and can be referenced by downstream nodes. A node's output Output states:
  • Success with data: The node completed and returned data
  • Success empty: The node completed but found no data (e.g., a search with no results)
  • Error: The node failed—an error message explains what went wrong

​
Mapping data between nodes

To use data from one node in another, you’ll use expressions. In any input field, you can reference outputs from previous nodes: 
{{input.company_domain}}
{{enrich_company.name}}
{{branch.yes.data}}

​
Connecting nodes

​
Creating connections

To connect two nodes, click on the output symbol of the first node and drag the connection to the input symbol of a subsequent node. Connect nodes in an execution flow

​
Deleting connections

To delete a connection, hover over the middle of the connecting line between two nodes and press the delete key. Delete a node connection

​
Connection rules

  • A node’s output can connect to multiple downstream nodes (fan-out)
  • A node can receive connections from multiple upstream nodes (fan-in)
  • Circular connections (loops) are not allowed
  • Branch and Switch nodes have multiple output ports for different conditions

​
Working with nodes

​
Adding nodes

Click the + button on the canvas or on a node’s output port to add a new node. Browse categories or search for specific actions.

​
Configuring nodes

Click on any node to open its configuration panel. Each node type has different settings, but common options include:
  • Node name: Rename the node for clarity
  • Input mappings: Connect data from previous nodes
  • Action settings: Configure the specific behavior
  • Error handling: Define what happens if the node fails

​
Organizing your canvas

  • Drag nodes to reposition them on the canvas
  • Zoom in/out using the scroll wheel or zoom controls
  • Select multiple nodes by clicking and dragging a selection box
  • Copy/paste nodes using keyboard shortcuts (Cmd/Ctrl + C, Cmd/Ctrl + V)

​
Testing tools

Before publishing, thoroughly test your tool:
  1. Click Test in the toolbar
  2. Enter sample input values
  3. Click Run to execute the tool
  4. Inspect each node’s output in the execution view
  5. Check for errors and unexpected results
Test mode uses real integrations and may consume credits or modify external systems. Use test data when possible.

​
Save and publish

​
Draft mode

When you first create a tool, it is automatically saved as a draft. You can leave and return later to finish building the tool without losing your progress. Publishing a draft

​
Publishing

Once you’re ready to start using the tool outside the editor, you can publish it. Published tools will be available for use in plays, agents and MCP servers. To publish a tool, click the Publish button in the top right corner of the tool editor.

​
Versioning

After publishing a tool, you can still make changes to it. Any changes you make will be saved as a new draft. You can publish the new draft at any time by clicking the publish button again.
Publishing creates a new version. Active plays and agents continue using the version they were configured with until you explicitly update them.

​
Best practices

Rename nodes from defaults like “Branch 1” to something meaningful like “Check if enterprise tier”. Your future self will thank you.
Don’t assume every node will succeed. Use Branch nodes to check for errors and handle them gracefully—log failures, send notifications, or trigger fallback logic.
Build tools that do one thing well rather than trying to handle every edge case. You can always chain tools together in plays.
Add descriptions to every input field. When someone else (or future you) uses this tool, they’ll know exactly what data to provide.
Don’t just test the happy path. What happens with missing data? Invalid formats? API rate limits? Test these scenarios before publishing.