Scripted manual testing / exploratory testing

Two ways humans test: following a written checklist step by step, or learning, designing, and running tests at once under a charter.

manual QA passjust click around and try to break ithuman testingmanual test scriptbug hunt sessiontest it by hand before we shipexploritory testingsomeone should actually use it first

See it

Live demo coming soon

What it is

Scripted manual testing is a human walking a written list: step, step, expected result, tick. Anyone can run it, two people get the same answer, and it produces an audit trail, which is why regulated and release-checklist work lives here. Exploratory testing is the opposite posture: learning the product, designing tests, and running them at the same time, with each result steering the next move. It is directed work, not wandering. A charter says what you are hunting, and what you observe decides where you go next. Cem Kaner named it; session-based test management (the Bach brothers) gives it structure with timeboxed sessions, a written charter, and notes.

Use both, for different jobs. Scripted passes verify that known behavior still holds. Exploratory sessions find the bugs nobody thought to write a step for: the double submit, the back button mid-payment, the 400-character name, the tab left open overnight. New features and risky refactors deserve an exploratory hour before they deserve a checklist.

Gotcha: exploratory testing is not 'unprepared clicking'. Without a charter ('30 minutes on file upload edge cases') and notes as you go, you cannot say what you covered or reproduce what you broke. And any scripted pass a human has run five times is a script the human now skims: automate it and give the person back to exploration.

Ask AI for it

Write two QA artifacts for the new invoice editor. First, a scripted manual test pass: numbered steps with preconditions, exact actions, and expected result per step, covering the happy path and the three main error states, formatted as a table a non-engineer can follow. Second, four exploratory testing charters in the form 'explore X with Y to discover Z', each timeboxed to 30 minutes, aimed at edge cases the script cannot cover: interrupted flows, unusual input lengths, permissions, and stale or concurrent sessions. Include a notes template for recording what was covered and what looked suspicious.

You might have meant

reproduction steps minimal reproducible exampleacceptance criteriatest casesmoke testtest plan