When testing, workflows will execute using the parameter values set for input variables.

Testing is a crucial part of building reliable AI agents. Kura provides two flexible approaches to test your workflows during development.

​
End-to-End Testing

The simplest way to test your entire workflow is using the “Run from Start” button in the builder. This executes your agent from the beginning and runs through all blocks sequentially.

This approach is ideal for:

  • Initial workflow validation
  • Final testing before deployment
  • Understanding the complete user journey

​
Selective Block Testing

For more granular testing, you can select specific blocks or sequences to test:

  1. Click on any workflow block to select it
  2. Hold Shift and click another block to select all blocks in between
  3. Click “Run Selected” to execute just those blocks

This targeted testing approach is particularly valuable when:

  • Debugging complex interactions that need multiple test iterations
  • Validating different branching scenarios in your workflow
  • Fine-tuning specific parts of your agent’s behavior
  • Testing error handling in specific sections

Pro Tip: Use selective block testing to rapidly iterate on challenging parts of your workflow without having to run through the entire agent sequence each time.

​
Debugging

When testing your workflow, you’ll notice an “Output” button that appears after executing a block. This opens up a panel that shows the output of each workflow block.

The output provides valuable insights including:

  • A clear explanation of what the agent did
  • The parsed HTML that the agent saw
  • The amount of tokens used by the agent
  • Whether the agent used cached information or made fresh requests

This visibility makes it much easier to:

  • Identify where things might be going wrong
  • Understand your agent’s decision-making process
  • Confirm that your agent is interpreting data correctly