Field Guide · Prompt Engineering for QA

Become AI-ready as a QA in one day.

One reusable prompt frame at 9 AM, then eight copy-paste recipes until your AI writes QA work you'd actually merge — strategy, plans, cases, data, API tests, automation code, bug reports, and refactors.

Day

Prompt Recipes

Framework

Blank Pages

// 09:00 · Warm-up

The R·C·T·F Frame

Every recipe below builds on these four moves. Learn it once — reuse it forever.

Role

Tell the model who to be. A persona sets vocabulary, rigor, and defaults.

"Act as a senior SDET with 10 years in test automation."

Context

Feed the facts: product, stack, constraints, and the source material to work from.

"App: checkout flow. Stack: React + Node. Here's the story + ACs: …"

Task

One verb, one deliverable. Be explicit about coverage and depth.

"Generate test cases for happy, negative, boundary, and edge paths."

Format

Pin the shape of the answer so it drops straight into your tools.

"Output as a markdown table: ID, steps, data, expected, priority."

+ Power-ups

Examples — paste 1–2 of your real cases so it matches your house style. Constraints — "no hard waits, max 12 cases". Iterate — "now add negative cases" beats rewriting from scratch.

// The Day

09:00 → 17:00 at a glance

Morning designs the tests. Afternoon builds and ships them.

09Strategy

10Test Plan

11Test Cases

12Test Data

13Lunch

14API Tests

15Automation

16Bug Reports

17Review

Morning · Design the Tests09:00 – 12:00 · think before you automate

09:00STRATEGY01

Test Strategy

Produce a risk-based strategy for the whole release — scope, levels, tooling.

strategy.md

Act as a QA lead. We're building [product/feature] for [users].
Stack: [stack]. Top risks: [risk areas]. Release: [cadence].

Draft a test strategy: scope, test levels (unit → E2E),
types (functional, perf, security, a11y), environments,
entry/exit criteria, tooling, and risk-based priorities.

Format with headings + a risk matrix table.

TIPStart broad, then ask "which risks deserve automation first?"

10:00PLAN02

Test Plan

Produce a sprint-ready plan with scope, schedule, deliverables, and owners.

testplan.md

Act as a senior QA engineer. Feature: [feature].
Acceptance criteria: [paste ACs].
Sprint [sprint], target release [date].

Create a test plan: objectives, in/out of scope,
deliverables, schedule, resources, environments,
test data needs, and a RACI.

Output as markdown with a milestone table.

TIPPaste the real ACs — generic plans come from generic context.

11:00CASES03

Test Cases

Produce detailed cases across happy, negative, boundary and edge paths.

cases.md

Act as a QA analyst. From this story + ACs:
"[paste story + acceptance criteria]"

Generate test cases for happy path, negative,
boundary, and edge cases. For each give: ID, title,
preconditions, steps, test data, expected result, priority.

Output as a markdown table.

TIPFollow up: "list cases you'd skip and why" to surface gaps.

12:00DATA04

Test Data Generation

Produce realistic datasets — valid, boundary, and invalid, ready to load.

data.csv

Generate [N] rows of realistic test data for
[entity] with fields: [field:type, …].

Include valid, boundary, and invalid records, plus
locale and edge cases (nulls, max length, special chars).

Output as CSV + a JSON sample, and
flag which rows are negative cases.

TIPAdd "PII-safe, synthetic only" so nothing real leaks in.

Afternoon · Build & Ship14:00 – 17:00 · turn cases into running code

14:00API05

API Test Scaffolding

Produce runnable API tests from a contract — status, schema, auth.

api.spec

Act as an SDET. Here's the API contract:
[paste OpenAPI / endpoint].

Write tests for [endpoint] covering 2xx/4xx/5xx,
schema validation, auth, and required/optional fields.
Use [pytest+requests / RestAssured / Postman].

Include positive, negative, and contract tests.
Output complete runnable code.

TIPHand it the spec file — it'll infer schemas you'd forget.

15:00AUTOMATION06

Automation Code

Produce Playwright / Selenium tests using the Page Object Model.

e2e.ts

Act as an automation engineer. Convert these
manual cases: "[paste cases]" into
[Playwright (TS) / Selenium (Java)] tests using the
Page Object Model.

Include locators, assertions, setup/teardown, and
data-driven params. Add comments.

Output complete, runnable files.

TIP"Use role/test-id locators, no XPath, no hard waits."

16:00BUGS07

Bug Reports

Produce clean, triage-ready reports from messy repro notes.

bug.md

Turn these raw notes into a bug report:
"[paste notes / logs]".

Include: title, severity, priority, environment,
preconditions, steps to reproduce, expected vs actual,
and an attachments checklist.

Suggest a likely root-cause area.
Format for [Jira / Azure DevOps].

TIPAsk for a one-line title a triager can scan in 2 seconds.

17:00REVIEW08

Review & Refactor

Produce a reliability review — flaky waits, brittle locators, fixes.

review.diff

Review this test code: "[paste code]".

Flag flakiness, hard waits, brittle locators,
missing assertions, and duplication.

Refactor for reliability and readability
(POM, explicit waits, fixtures). Explain each change briefly.

Output the improved code + a review checklist.

TIPEnd the day here — clean tests today save triage tomorrow.

/ GOLDEN RULE 01

Context beats cleverness.

Paste real stories, code, and logs — the model is only as sharp as what you feed it.

/ GOLDEN RULE 02

Never ship unread.

AI drafts; you verify. Run it, check the assertions, own the merge.

Save it. Share it. Use it tomorrow.

New AI-for-QA guides drop here every few weeks. If this was useful, send it to a fellow QA engineer who's wrestling with their test backlog.

All AI for QA guides Browse free resources

By Sreenidhi Rajakrishnan · Automation Architect · AI & Automation