Skip to content

Test Authoring Templates

A Test Authoring Template is a saved run configuration. It’s the starting point for testing in OttoTester — you pick a template, point it at a variant, and run it.

The template tells the agent pipeline what to do: which phases to run, how many test cases to aim for, which categories of test to cover, and how to execute. Save it once, reuse it every run. Templates belong to an application, and one is marked the default.

The most important field is optional: the Agent Prompt.

OttoTester explores your app on its own. The Agent Prompt is how you steer it. It’s a free-text box — plain instructions for the agents, in your own words:

  • “The save button takes 2–3 seconds — wait for the toast.”
  • “Focus on the checkout flow.”
  • “The date fields use a custom picker, not a native input.”

Leave it empty and OttoTester tests what it finds. Fill it in and the agents weight their work toward what you said. You’re never writing a script — you’re giving the agents context and direction.

If your application has candidate tests, the template shows them as a tickable list under the Agent Prompt. Tick the ones you want and each becomes its own test, linked back to the candidate it came from — so the candidate’s “tests created” count is exact.

Picked tests are created exactly as selected; they don’t go through the planner’s wording, so the link can’t get lost. The Agent Prompt box is then only for extra tests beyond the ones you ticked — leave it empty to create only your picks, or describe more tests to have OttoTester add them on top.

A template runs in one of two modes:

  • Discover & generate — the normal mode. The planner explores the app, picks new test cases, and the pipeline generates and runs them.
  • Resume incomplete work — picks up test cases that earlier runs left unfinished (still planned, blocked, or scripted with a failing script) and processes them in batches. Use it to chip away at a backlog without re-planning from scratch.

Beyond the Agent Prompt and mode, a template bundles:

  • Phases — which pipeline steps run. In discover mode, plan is always on; generate, execute, and heal are toggles. Resume mode skips planning entirely.
  • Test categories — what kind of testing to cover. Search is the supported category today; create/edit/delete testing is coming but locked for now, since those actions can change your app’s data.
  • Max test cases — how many to aim for in one run.
  • Feature status — tells the healer how aggressively to repair. Existing feature → repair brittle tests aggressively. New or changing feature → stay conservative, because a failure might be a real behavior change.
  • Execution settings — page timeout, headed vs headless, screenshots, trace capture, log verbosity, and whether to use test data.