Using the Scout UI

Scout comes with an integrated web interface (UI) built with React, providing a user-friendly way to manage and monitor your Tests. This guide will walk you through the main sections and functionalities of the Scout UI.

(Note: As the UI evolves, some visual details or layouts might change. This guide focuses on the core functionalities accessible through the interface.)

Assuming Scout is running (see Getting Started), you can typically access the UI by navigating to http://localhost:3000 in your web browser.

Dashboard Overview

Upon logging in or accessing the main page, you are typically greeted by a dashboard. This area provides a high-level overview of your Scout instance and active Tests.

Key information and actions on the Dashboard include:

  • List of Active Tests: A summary table or list displaying all current Tests.

    • For each Test, common details shown include its name, creation date, number of variants, and potentially top-level performance indicators (e.g., total predictions, total updates).

  • Create New Test Button: A prominent button or link to navigate to the Test creation form.

  • System Status Indicators: Basic health indicators for connected services like Redis.

  • Navigation Menu: Links to other sections of the UI, such as Admin Controls, Log Viewer, and documentation.

(TODO: Add a screenshot of the main dashboard if available)

Creating a New Test

One of the primary functions of the UI is to facilitate the creation of new Tests.

Steps to create a new Test typically involve:

  1. Navigate to the "Create Test" or "New Test" section (often from the dashboard).

  2. Fill out a form with the following details:

    • Test Name: A descriptive name for your Test (e.g., HomePageHeroButtonTest, ArticleRecommendationStrategy). This name is for your reference.

    • Variants: Define the different options you want to test.

      • You will usually be able to add multiple variants.

      • For each variant, you will provide a Label (e.g., "Red Button", "Learn More Link", "Algorithm V2"). Scout will assign an internal ID to each variant (typically starting from 0).

    • The UI may provide options or guidance on setting up Tests that utilize contextual data.

  3. Submit the form.

Upon successful creation, Scout will:

  • Register the new Test with the backend.

  • Assign it a unique cb_model_id (Test ID).

  • The new Test will appear in your list of active Tests on the dashboard.

(TODO: Add a screenshot of the test creation form if available)

Managing Existing Tests

Once Tests are created, you can manage them through the UI.

Actions available for existing Tests include (usually by selecting a Test from the dashboard list):

  • View Details / Performance: Clicking on a Test name or a "details" button will take you to a dedicated page for that Test (see next section).

  • Delete Test: An option to permanently delete a Test and all its associated data. This action is usually irreversible and should be used with caution.

    • Corresponds to the POST /api/delete_model/{cb_model_id} API endpoint.

  • Rollout Global Variant (Override): For a specific Test, the UI may allow you to temporarily or permanently force all recommendations to a single, specific variant. This is useful for fully rolling out a winner or for incident management.

    • Corresponds to POST /api/rollout_global_variant/{cb_model_id}.

  • Clear Global Variant Rollout: If a global variant rollout is active, an option to clear it and return to normal Test operation.

    • Corresponds to POST /api/clear_global_variant/{cb_model_id}.

Viewing Test Performance & Details

Selecting a specific Test provides detailed insights into its operation and performance.

Information and visualizations you can find on a Test detail page:

  • Test Information: Name, cb_model_id, creation date, list of variants (ID and label).

  • Performance Metrics (often with charts):

    • Reward Trends: Average reward over time for each variant and overall.

    • Variant Performance: Cumulative rewards, number of times each variant was chosen (predictions), number of updates received.

    • Prediction Ratios: The proportion of times each variant has been recommended.

    • Contextual Feature Insights: (For Tests using context, if available) Information about how contextual features might be influencing decisions.

  • Recent Activity/Trail: A log or table showing recent prediction requests and updates for this specific Test (timestamps, chosen variants, rewards, context snippets if applicable).

  • Configuration Details: Information about the Test's configuration (though deep configuration of the underlying statistical methods might be limited via the UI, which typically favors robust defaults).

(TODO: Add a screenshot of a test detail/performance page if available)

Admin Controls

The Admin section of the UI allows for managing system-wide settings and security.

Key functionalities in the Admin Controls area:

  • API Protection:

    • View current status (whether API protection is enabled or disabled).

    • Toggle API protection on/off (POST /admin/set_protection).

  • Authentication Token:

    • Generate a new authentication token (POST /admin/generate_token). The new token will be displayed for you to copy and use in your API requests if protection is enabled.

    • (Note: The UI might not display the current token for security reasons after its initial generation, only allow generating a new one).

  • Configuration Viewer/Editor:

    • View current system configuration (GET /admin/system_config), Test defaults (GET /admin/model_config), and Redis settings (GET /admin/redis_health might show connection status; GET /admin/system_config may include some Redis parameters).

    • Possibly edit some of these configurations (POST /admin/model_config, POST /admin/system_config, POST /admin/redis_config).

  • Redis Health Check: An indicator or button to check the connectivity and health of the Redis instance (GET /admin/redis_health).

(TODO: Add a screenshot of the admin controls section if available)

Log Viewer

Scout includes a real-time log streaming feature in the UI.

  • Live Log Stream: Displays logs from the backend application as they happen.

    • Corresponds to the GET /logs/stream API endpoint.

  • Filtering/Search (Optional): The UI might offer basic filtering or search capabilities for the logs.

This is extremely useful for debugging, monitoring requests in real-time, and understanding system behavior without needing to access server-side container logs directly.

By familiarizing yourself with these sections, you can effectively leverage the Scout UI to manage your Tests and gain valuable insights into their performance.

PreviousLicenseNextScout Kubernetes Deployment

Last updated