*Note: This feature is subject to AI credits. AI credits are required to use this feature.
Table of Contents
- MagicPod Autopilot Overview
- Main Features and Characteristics
- Usage Instructions
- Using shared steps
- Information passed to LLM(Generative AI)
- Regarding conversation history
- Planned Feature Expansions
- Common Issues and Solutions
1. MagicPod Autopilot Overview
What is MagicPod Autopilot
MagicPod Autopilot is a test automation agent that leverages AI technology. It enables test case creation and editing through natural language input. It simplifies the traditional test creation process, allowing both technical and non-technical users to create test cases intuitively.
Supported Platforms
- Mobile Applications: iOS, Android
- Web Browsers: Web application testing across various browsers
2. Main Features and Characteristics
Test Step Generation Using Natural Language
The key feature of MagicPod Autopilot is its ability to automatically generate executable test steps simply by describing test content in natural language. For example, from English instructions like "Enter username and password, then click the login button," corresponding test steps are generated.
Flexible Test Case Editing Capabilities
In addition to adding steps when creating new test cases, editing existing test steps is now possible. Steps can be added or modified at any position within test cases, enabling more efficient maintenance of existing test cases.
Autonomous Step Verification
The system includes functionality to automatically execute created test steps and perform operation verification. AI verifies whether the generated steps work as expected, supporting improved test quality.
Test Case Analysis and Summarization
By loading existing test cases, AI can analyze their content and generate explanations and summaries. This streamlines understanding of complex test cases and facilitates information sharing among team members.
Failure analysis feature
The failure analysis feature is a capability in which MagicPod Autopilot automatically investigates and analyzes the root cause when a test fails.
For details, please refer to the following:
About the failure analysis feature
3. Usage Instructions
Prerequisites
To use MagicPod Autopilot, the "Generative AI" feature must be enabled by an organization administrator account. (Refs: How to Enable Generative AI Features)
Test Step Generation Procedure
- Open the MagicPod test case editing screen
- Click the MagicPod Autopilot button at the bottom right of the screen
- Enter the test content you want MagicPod Autopilot to do in natural language
- Example: "Add steps to enter email address and password in the login form and press the login button"
- Example: "Please briefly explain what kind of processing is involved in steps 23-41."
- Click the bottom right button or use keyboard shortcut (Command+Return / Ctrl+Enter)
-
AI processing begins on MagicPod Autopilot's chat interface
-
The response completes after a while. (When the test case is updated, the steps on the screen are automatically refreshed)
4. Using shared steps
When creating test steps using MagicPod Autopilot, you can use existing shared steps or create new shared steps. Please note the following points:
- Editing existing shared steps is not supported.
-
If you want to use or create shared steps, specify this clearly in your prompt.
Prompt examples
- Create a test case for user registration. Please use any available shared steps.
-
Create the user registration steps as a shared step.
-
When using an existing shared step, its name and description are referenced. We recommend setting an easy-to-understand name and a detailed description.
Note that you can update the shared step name and description in the "Information" tab of each shared step.
5. Information passed to LLM(Generative AI)
- Prompts entered by users
- Test steps
- UI images and UI tree information
- Only currently displayed UI. Past UI images are excluded.
- Data patterns
- Shared variables
- Excluding secret shared variable values
- Line numbers and error messages where test failures occurred
6. Regarding conversation history
7. Planned Feature Expansions
The following features are planned for future releases:
- Generative AI model selection feature: Allow changing from the default model to a different one (e.g., changing from GPT-5 to Claude Sonnet 4).
- File upload functionality: Enable passing files such as Excel specification documents to MagicPod Autopilot to create test cases based on the file contents.
- Test Self-healing on Semantic Context: By applying MagicPod Autopilot's test creation engine to the AI self-healing functionality already built into MagicPod, it will enable more advanced repair that understands user intent.
8. Common Issues and Solutions
The button to launch MagicPod Autopilot is not displayed
The launch button will not appear if the "Generative AI" feature is not enabled.
Please request the organization administrator to enable the "Generative AI" feature
How to check the AI credit balance
Please refer to this page
The error "The request exceeds the model's maximum context length. Please start a new chat to continue." occurs
This error occurs when the input text sent to MagicPod Autopilot is too long, and the LLM model reaches its processing limit.
The input text includes the following:
- The prompt sent by the user to Autopilot
- The responses returned from Autopilot
- Information from the loaded test cases
In other words, the input includes the entire conversation history as well as the contents of any referenced test cases. As a result, if you are working with lengthy test cases, you may reach the limit sooner.
Solution:
We recommend starting a new chat as frequently as possible.
As conversations grow longer, not only does AI credit consumption increase, but response accuracy tends to decrease as well. Please reset your chat at appropriate intervals.