Documentation

How to test with Cettara — by hand, automatically, or through an AI assistant.

What Cettara is

What Cettara isOne place for manual and automated testing, where every automated verdict shows its reasoning.
Quick startFrom signing up to your first recorded result, by hand or automated.
ConceptsThe words used throughout Cettara, and what each one means.

Manual testingCases people follow, runs people record, and the evidence both leave behind.
Writing a manual caseSteps, expected results, preconditions and the three kinds of step.
Running a manual testThe run wizard, the four outcomes, and what is recorded.
Assignments and My workWho is meant to run what, and what is waiting for you.

AutomationCases a browser runs, with an AI judge ruling on what a person would have seen.
Setting upEnvironments, variables, generated values, app adapters and fixtures.
Writing a test caseEvery kind of step, and when to use it.
Checkpoints and Do stepsThe two steps that use AI — how to write them, and what they cost.
Blocks and flowsWrite steps once, and group cases that belong together.
Running and reading a runStarting a run, watching it, and what it leaves behind.
The AI judgeHow a verdict is made, how to correct it, and how it learns.

Evidence and sign-offWhat a run keeps, how long, who signs it, and the audit log behind all of it.

AI assistants (MCP)Let Claude, Claude Code or Cursor read and write your test cases through the Model Context Protocol.
Connect with OAuthOne click from the Claude apps — no token to copy or store.
Connect with an API tokenFor Claude Code, Cursor and scripts — a token you create and paste once.
Tools referenceEvery tool an assistant can call, and the permission each one needs.
Permissions and limitsWhat an assistant is allowed to do, and what stops it.