How to use AI code assistants with Squish

General Guidance

General guidance on Squish test script generation with AI code assistants, such as GitHub Copilot, Cursor, or Claude CLI, is given in the blog post A Practical Guide to Generating Squish Test Scripts with AI Assistants.

The guidance given especially considers:

• Context
• Fast iteration or feedback loop from test results to improvements.

Context includes everything from up-to-date Squish documentation to team-specific rules for test scripts.

Establishing a connection from the AI code assistant to running the test scripts and receiving the test results is needed to make the test script iteration faster. With the test results, the AI code assistant is able to fix errors autonomously.

For creating the connection between Squish and the AI code assistant use Squish Model Context Protocol (MCP) with your AI code assistant.

Squish MCP enables MCP-compatible AI agents to interact with Squish and the application under test (AUT). It gives the agent access to the current state of the running application and provides tools for performing UI interactions through Squish. This allows the agent to explore the AUT, generate or work with existing test cases, execute them, analyze results, and refine test scripts.

Use Squish MCP when you want the AI agent to work autonomously, for example to:

• Generate new Squish test cases from a high-level testing intent.
• Explore application workflows and UI states.
• Run generated or existing tests.
• Analyze whether a failure is caused by the application or by the test.
• Improve generated or existing test scripts based on execution results.

To get started with the Squish MCP 0.2.0 technology preview, download the package from the Qt Customer Portal. The package includes offline documentation with setup instructions, including guidance for configuring Squish MCP in VS Code.