What's New: Running Tests from Visual Studio Code (2026 release wave 1)
Running AL tests from Visual Studio Code in the 2026 release wave 1: test explorer, run profiles, code lens, debugging, code coverage and Copilot agent test fixing. It also covers limits: no AI or data-driven test suites, because AL test runners are not used.
- Overview of Testing Capabilities
- Testing Explorer and Code Coverage Demo
- Debug Tests in VS Code
- Copilot Agent for Test Fixing
- Technical Implementation Details
- Agentic Development Results
- Related Resources and Conclusion
Overview
The video shows the added ability to run tests in Visual Studio Code. A testing explorer lists tests grouped by project and codeunit with pass/fail status. Run profiles cover publish and run, run only, debug, and run with code coverage. Code lens shows how many tests have executed a procedure and lets you run the associated tests from the code. Debugging a test sets up the debug session automatically and can be started from the test explorer, the test sidebar, or test results.
It also covers a Copilot agent that can run tests, find failing ones and fix the test code, with several runs to validate the fixes. The technical section explains that the framework uses launch.json, configures test isolation dynamically, and does not use AL test runners. The AL MCP tool, covered in a separate session, serves agentic tools outside VS Code.
Key points
- Tests run directly in VS Code, with a testing explorer grouping tests by project and codeunit and showing pass/fail status; this uses only platform parts, with no extra application added.
- Run profiles: publish and run in one operation, run only, debug, and run with code coverage collection.
- After a code coverage run, code lens shows how many tests executed a procedure and how many passed, and clicking it runs those tests.
- Debugging can start from the test explorer, the test sidebar, or test results (‘debug failed test’), and the debug session is set up automatically.
- Copilot agents can run tests out of the box once the run test tool is enabled; in the demo the agent fixed the bug and ran the tests again to validate the fix.
- The framework uses launch.json, so it can target multiple sandbox or on-prem servers and pick the test company through the startup company property.
- AL test runners are not used, so AI and data-driven test suites are not supported, and codeunits relying on test runner setup and teardown events may not work.
Features
| Feature | Status | At |
|---|---|---|
| Run tests in Visual Studio Code | status not stated, demoed | 0:17 |
| Run profiles for tests | status not stated, demoed | 0:35 |
| Code lens for test execution | status not stated, demoed | 2:08 |
| Debug tests from VS Code | status not stated, demoed | 3:16 |
| Copilot agent for test fixing | status not stated, demoed | 4:26 |
| Test framework using launch.json configuration | status not stated | 5:01 |
| No support for AI and data-driven test suites | status not stated | 5:37 |
| Dynamic test isolation configuration | status not stated | 5:52 |
| Disable test integration in VS Code | status not stated | 6:18 |
| AL MCP for agents outside VS Code | status not stated | 7:46 |
Quotes
- 0:17 “In this release, we have added the ability to run tests in Visual Studio Code.”
- 0:35 “Part of this is a set of run profiles. There is the profile for publishing and running your tests.”
- 3:16 “But now, from VS Code, it’s like really, really easy to set up a run and debug.”
- 4:37 “the good thing about having integrated with the VS Code tools is that we get the ability for agents to run tests out of”
- 5:37 “we are not using AL test runners. This means that we currently do not have any support for running AI or data-driven test suites.”
- 6:06 “we actually configure this dynamically based on the required test isolation property that’s set on the test code unit.”
- 6:18 “you can disable the test integration in VS Code using the AL disable test running setting.”
- 6:44 “giving the agent the ability to run and validate its own AL test own AL code using tests um significantly improves the accuracy of”
- 7:46 “we will have a bit on the AL MCP. And this is a tool for those of you who are not using VS Code”
Disclaimers in the video
- 5:37 not-in-this-release: we currently do not have any support for running AI or data-driven test suites
Presenters (as heard): Thomas, Thaddeus.