Teststream documentation

Teststream runs browser tests against your application. Tests live in Teststream; you can run them from a terminal or pipeline, report results on a pull request, and let a coding agent propose changes for review.

Start here for CLI, CI and coding-agent workflows. Test-writing syntax is available as supporting reference.

Choose your workflow

Goal	Guide
Connect a terminal to Teststream	Set up the CLI: install from npm, API key, project and connection check.
Make CI pass or fail on test results	Run tests in CI: selection, environments, exit codes, timeouts and JSON artifacts.
Report on GitHub or GitLab	Test a pull or merge request: runnable CI examples, preview URLs, affected tests and merge checks.
Decide what to do with a failure	Investigate failures: status, report, diagnose and reviewed propose-fixes.
Give an agent access to tests	Use a coding agent: install skills, configure MCP and verify branch proposals.

Reference and explanation

CLI commands lists the available commands, options, output formats and exit codes. How Teststream runs tests explains where execution happens, replay versus AI, healing, and the difference between a passing run and a reviewed test change.

The usual path is:

Deploy the app -> run tests -> inspect evidence -> diagnose failures
                                             -> fix the app, setup or test
                                             -> verify -> review

Teststream does not build or deploy your application. AI execution, diagnosis and drafting can incur model costs; replay avoids model calls for the recorded actions.

Writing tests

When you need to create or update a test, use Write a test and the test file format. A first-test tutorial is also available.