Skip to content

Start a New Test Run

This guide will walk you through starting a new test run using the Litmus UI.

Prerequisites:

  • You have deployed the Litmus application.
  • You have access to the Litmus UI in your browser.
  • You have created at least one Test Template.

Steps:

  1. Navigate to the "Start New Run" page:

    Start new run

    • From the Litmus UI, click on the "Start New Run" item in the left sidebar.
  2. Select a Test Template:

    • Choose the desired template from the "Template ID" dropdown. This will display available templates of both types: "Test Run" and "Test Mission".
    • The available test cases, input, and output fields for the chosen template will be shown.
    • When selecting a "Test Mission" template, you'll see the mission duration defined in the template.
  3. Enter a Run ID:

    • Input a unique identifier for this specific test run in the "Run ID" field.
    • This ID will help you distinguish this run from others.
  4. (Optional) Modify Request Payload:

    • Review the default request payload displayed in the "Request Payload" tab.
    • You can modify this payload if needed, using the available tokens: {query}, {response}, {filter}, {source}, {block}, {category}. For "Test Missions", the {query} token will be replaced by the LLM's generated requests in each interaction loop.
    • Ensure the payload is valid JSON.
  5. (Optional) Configure Pre-Request and Post-Request:

    • If required, you can define a pre-request to be executed before each test case in the "Pre-Request" tab.
    • Similarly, you can define a post-request to be executed after each test case in the "Post-Request" tab.
    • These requests are optional and allow you to set up and tear down your testing environment.
  6. (Optional) Test Your Request:

    • Before submitting your test run, click the "Test Request" button within the "Request Payload" tab.
    • This sends a single request using the defined payload and allows you to verify it's correctly formatted and reaches your endpoint.
    • If successful, you can then select the "Output" tab to view the response structure.
  7. Submit the Test Run:

    • After reviewing all configurations, click the "Submit Run" button.
    • This initiates the test run using the chosen template and your configurations.
    • If you chose a "Test Mission" template, the LLM will be invoked to guide the interactions for the specified duration.
  8. Monitor Run Progress:

    • You can track the progress of your run on the "Tests" page, accessible from the sidebar.
    • The page displays the run ID, start time, status, and progress. For "Test Missions", the progress shows the current turn in the mission.
    • When the run completes, its status will change from "Running" to "Completed"
  9. View Results:

  • Click on a completed run ID to view its detailed results.
  • For "Test Runs", this page displays each test case and its outcome: success or failure, along with assessment results.
  • For "Test Missions", you'll see the full conversation history, the final assessment, and details about each interaction loop.
  • You can further explore individual requests, responses, and assessment results (if applicable) by clicking the "Explore" button.

Note:

  • Ensure that the request payload and pre/post requests (if defined) are valid JSON structures.
  • If using the "Test Request" function, remember to select the output field for viewing the response structure.
  • You can use filter options to narrow down the displayed test cases on the run results page.

That's it! You've successfully started a new test run.

Disclaimer: This is not an officially supported Google product.