Turn a requirement into a reviewable test plan
Paste a user story or requirement and get test scenarios that each point back to your text, plus the questions it leaves open.
Paste a requirement, user story, or acceptance criteria. You get a draft test plan: happy paths, boundaries, failure cases, security, accessibility, and compatibility, where relevant. Every scenario either quotes the line of your requirement it tests or is labelled as an inferred risk with its assumption spelled out. Vague, missing, or contradictory requirements come back as open questions instead of guesses. Review, edit, and pick the scenarios you want, then export Markdown or JSON. It's planning help: nothing is executed.
What to paste
A requirement or user story, optionally with acceptance criteria, product context, and browsers or platforms in scope.
What you get
Test scenarios with steps and expected results, each traced to your text or labelled as an assumption, plus open questions.
Try it
How AI is used. This tool drafts a proposal from your text. It can be wrong or incomplete. Items marked “From your text” quote where they came from; “Inferred” items are the AI's reasoning; “Needs your input” means your text didn't say.
Your text. Your text is sent to our server to build the plan and isn't stored or logged. When live generation is on, it's also sent to an AI provider (Anthropic) to draft the plan. Don't paste secrets or personal data.
Optional details: acceptance criteria, title, context, platforms
Paste your text or load the example to begin.
Worked example
A synthetic example you can run above with “Load the password-reset example”. It works even when live generation is off, and its result is labelled as a prepared sample.
As a registered user who has forgotten my password, I want to reset it through a link sent to my email so that I can sign in again.
How it works
- Your requirement is split into numbered source lines on our server before anything else happens, and every scenario must point to those numbers or say it's an inferred risk.
- The AI's answer has to match a versioned schema. Scenarios that cite text you didn't write, or claim that tests ran or passed, are removed, and you're told what was removed.
- Ids are assigned by the server, not the AI, so they stay stable and sequential.
- Every change to the tool runs a fixed set of normal, vague, contradictory, and adversarial requirements through the same checks before it can go live.
How your text is handled: privacy. Terms for AI results: terms.
Limitations
- It plans tests; it doesn't run them. Nothing here shows that any system was tested or works.
- It only knows what you paste: no links are followed and no code or repository is read.
- Inferred scenarios are suggestions based on common risks, not requirements. Check each assumption.
- Live generation isn't switched on yet: while this tool is experimental, only the example runs.
Built by Ali Safari at ASafariM Digital · Page last reviewed