50 ChatGPT Prompts for Software Testers (2026): Manual, Automation, API

Last updated: July 2026

This is the working library the SoftwareTestPilot team uses every day. 50 ChatGPT prompts for software testers, grouped into 8 clear buckets — manual test cases, bug reports, API, automation code, SQL/database, performance/security/accessibility, test plans/strategy, and interview/career. Every prompt has a use-case, a copy button, and an example of what good output looks like. At the end you'll find 10 advanced prompts that use Custom GPTs, and honest notes on when GPT-4o beats GPT-5 (and vice versa).

Pair this with our Claude AI for test case generation guide and the GitHub Copilot for QA tutorial — most senior testers keep all three tabs open and pick the right tool per task.

How to use this page: click the copy button on any prompt block, replace the [bracketed] parts with your real feature/product context, paste into ChatGPT, then score the output using the 6-point rubric further down. Never accept the first draft.

Every strong prompt below follows the same 4-part structure. Learn it and you can invent your own prompt for any feature.

If output feels generic, one of the four parts is missing. Add it and re-run.

Rule of thumb: GPT-4o for volume and vision, GPT-5 for reasoning and code. If you have Plus, keep both pinned.

Use these to turn a user story or acceptance criteria into a reviewable test-case table. Every prompt asks for a markdown table you can paste into Jira/TestRail/Xray.

Prompt 1 — Login flow

Act as a senior QA engineer. Create positive, negative, boundary, and
security-focused test cases for a login page with email, password, remember-me,
forgot-password, account lock after 5 failed attempts, and CAPTCHA after 3
failed attempts. Return a markdown table with TC ID, Priority, Preconditions,
Steps, Test Data, Expected Result, Type.

Prompt 2 — Signup + verification

Act as a senior QA engineer. Write detailed test cases for a signup form
collecting name, email, phone, password, referral code, and terms acceptance.
Cover validation messages, duplicate-user, mobile-number formats (India + US +
UK), password strength rules, and email-verification link expiry.
Output: markdown table with the columns above.

Prompt 3 — Ecommerce checkout

Act as a senior QA engineer for a B2C ecommerce app. Generate end-to-end test
cases for checkout covering: address selection, single-coupon rule, wallet
balance, card + COD + UPI payment, tax + shipping, failed-payment retry, and
order-confirmation email. Include 5 explicit negative cases and 3 boundary
cases. Markdown table.

Prompt 4 — Search + autocomplete

Create test scenarios for a product-search feature with autocomplete, spelling
correction, filters, sorting, pagination, and no-result state. Include edge
cases for special characters, emoji, RTL text, and 200-character queries.
Markdown table with Priority and Type.

Prompt 5 — File upload

Create test cases for file upload supporting PDF, PNG, JPG up to 5MB. Cover
invalid extension, corrupted file, duplicate upload, slow network (3G throttle),
preview, delete, and virus-scan pending status. Include 3 security cases
(polyglot files, path traversal in filename, XSS in filename).

Prompt 6 — Role-based admin

Generate test scenarios for admin user management: invite user, change role,
deactivate account, reset password, view audit logs. Include access-control
negatives (member calling admin endpoint directly, expired-token replay).
Markdown table.

Prompt 7 — Subscription billing

Create test cases for monthly + yearly subscription billing with trial period,
upgrade, downgrade, mid-cycle cancellation, invoice download, tax, card expiry,
and payment-failure retry. Add 3 dunning-flow cases.

Prompt 8 — Notification center

Write test scenarios for in-app notifications with read/unread, bulk
mark-as-read, delete, filters, real-time updates via websocket, and
email-notification preference. Include reconnection + missed-message replay.

Prompt 9

Rewrite this rough bug note into a Jira bug report with Title, Environment,
Steps to Reproduce, Actual Result, Expected Result, Severity, Priority,
Attachments Needed, Business Impact. Keep it under 250 words.

[paste your notes]

Prompt 10

I found: payment succeeds but order stays 'pending'. Ask me the 8 missing
questions needed to file a complete defect report. One question per line, most
important first.

Prompt 11

Given this issue [paste symptom + logs], list 5 possible root causes ranked by
probability. For each, suggest the exact log line, DB query, or metric that
confirms or eliminates it.

Prompt 12

Convert this developer explanation into 2 sentences of plain-language release
note copy that a customer can understand:

[paste explanation]

Prompt 13

Review this bug report. List: (1) missing evidence, (2) unclear repro steps,
(3) severity/priority mismatch, (4) sections that need screenshots or HAR.

[paste bug]

Prompt 14

Group these 30 bugs [paste titles] by likely shared root cause. Return a table
with cluster name, member IDs, confidence, and suggested owning team.

Prompt 15

Create API test cases for POST /orders with fields userId, items, couponCode,
addressId, paymentMode. Cover 200/400/401/403/404/409/422/429/500. Include
idempotency-key handling, pagination, and rate-limit assertions. Markdown
table: Case, Method, Path, Headers, Body, Expected Status, Assertions.

Prompt 16

Generate Postman test-script assertions for this response. Check status,
schema, required fields, data type, response time under 800ms, and business
rule that total = subtotal - discount + tax. Output as pm.test blocks.

[paste response JSON]

Prompt 17

Create boundary tests for an API accepting page, size, sortBy, sortDirection.
Include invalid values, missing values, max size, negative numbers, and
SQL-injection attempts in sortBy.

Prompt 18

Write contract-testing scenarios for a user-profile API consumed by web,
Android, iOS, and analytics services. Highlight fields each consumer relies on
and breaking-change risks.

Prompt 19

Given this OpenAPI snippet [paste], identify missing test cases, risky
assumptions, and any endpoint that lacks explicit 4xx documentation.

Prompt 20

Generate GraphQL security + coverage tests for this schema [paste]. Cover
depth limit, alias abuse, batching abuse, N+1 patterns, per-field auth, and
introspection in production.

Practice more with our API testing interview questions and Postman tutorial.

Prompt 21

Write a Playwright TypeScript test for login using role-based locators
(getByRole), env-vars for credentials, and assertions for URL and welcome
heading. No waitForTimeout, no page.pause. Use expect.poll for the
async-loaded profile menu.

Prompt 22

Convert these manual test steps into Cypress tests using data-testid selectors
and cy.intercept for network validation. Keep tests independent.

[paste steps]

Prompt 23

Refactor this Selenium Java test into Page Object Model with clear methods,
explicit WebDriverWait (10s), and no Thread.sleep. Add TestNG DataProvider for
the login variants.

[paste code]

Prompt 24

Review this automation test for flakiness risks (fixed waits, unstable
selectors, shared state, weak assertions). Rewrite the risky parts.

[paste code]

Prompt 25

Create a test-data factory in TypeScript for users with default values and
overrides. Show how to use it in Playwright tests via a fixture.

Prompt 26

Convert this Playwright user journey into a k6 load script. 50 VUs for 5 min,
p95 < 800ms, error rate < 1%. Add think time 1-3s between requests.

[paste journey]

Prompt 27

Rewrite this pytest + Selenium test as Playwright TypeScript. Keep the same
test IDs and assertions. Move any Thread.sleep to expect.poll.

[paste test]

Prompt 28

Write a GitHub Actions workflow: Node 20, npm ci, playwright install, run
tests sharded across 4 workers, upload HTML + Allure report as artifacts, fail
the job on any test failure.

Deepen with our Playwright complete guide and Selenium interview questions.

Prompt 29

Generate SQL (Postgres 15) to insert 10 orders across 3 users: 2 pending, 6
shipped, 2 refunded. Use a CTE for user IDs. Include realistic timestamps in
the last 30 days.

Prompt 30

Write 8 data-integrity check queries for tables users, orders, payments,
refunds. Cover orphaned rows, duplicate unique keys, cascade-delete gaps,
negative amounts, and FK violations. Postgres syntax.

Prompt 31

Review these 3 queries [paste]. List missing indexes, N+1 patterns, full-table
scans, and suggest EXPLAIN ANALYZE commands to verify.

Prompt 32

Given this migration [paste DDL], generate test cases covering:
backward-compatibility, rollback safety, data-loss risk, lock duration on a
50M-row table, and application read/write during migration.

Prompt 33

Write a reconciliation SQL that matches orders vs payments vs refunds for a
day and flags any mismatches, unrefunded cancellations, or double-charge
suspects. Output as a report-friendly SELECT.

Sharpen SQL fundamentals with our SQL interview questions.

Prompt 34

Create a performance-testing checklist for a checkout API before a festive
sale. Include response time SLOs, throughput targets, error rate, DB pool
size, cache hit ratio, and monitoring dashboards to watch during the ramp.

Prompt 35

Write a k6 load test. Stages: ramp to 50 VUs over 2 min, hold 5 min, ramp
down. Endpoint: POST /api/orders with random payloads. Thresholds: p95<800ms,
error rate<1%. Add checks and a summary handler that writes summary.html.

Prompt 36

Generate security test cases for the login + password-reset feature mapped to
OWASP Top 10 2021: A01 broken access, A02 crypto, A03 injection, A07 auth
failures, A08 integrity. One test per category with expected response.

Prompt 37

Suggest security test cases for forgot-password: token expiry, token reuse,
brute-force on the reset endpoint, email-enumeration timing, and
session-invalidation after reset.

Prompt 38

Create accessibility test scenarios for a modal dialog mapped to WCAG 2.2 AA:
keyboard navigation, focus trap, ARIA labels, screen-reader announcements
(NVDA + VoiceOver), color contrast 4.5:1, and Escape-to-close behavior.

Prompt 39

List mobile-usability test ideas for a signup form used on low-end Android
devices (2GB RAM, Android 11) with 3G network. Include tap-target size,
keyboard behavior, autofill, and slow-network resilience.

Prompt 40

Create a release test strategy for a web app launching a new payments module.
Include scope, risks (top 5), environments, regression approach, automation
coverage targets, entry criteria, exit criteria, and go/no-go checklist.

Prompt 41

From these 100 test cases [paste], select the top 15 for a 2-hour smoke
regression. Factor in these 5 recent production incidents [paste]. Justify
each pick in one sentence.

Prompt 42

Create a QA status report format for daily standup showing tested items,
blocked items, defects (by severity), risks, and next plan. Keep it under 6
lines.

Prompt 43

Suggest QA metrics that show real quality improvement without encouraging
testers to file low-value bugs. Include definition, formula, and cadence.

Prompt 44

Create a test closure report for a sprint release: scope, coverage, open
defects by severity, deferred risks, automation additions, and 3 lessons
learned.

Prompt 45

Create a UAT checklist for business users testing invoice generation. Use
business language (no tester jargon). Include a sign-off table with role,
name, date, and comments.

Prompt 46

Act as a QA interviewer at a product company. Ask me 10 scenario-based
questions on API testing one by one. Wait for my answer, score it 1-5 on
correctness, structure, and depth, then give one follow-up question.

Prompt 47

Rewrite this QA resume bullet to show impact, tools used, and a measurable
result. Keep it under 25 words. No buzzwords.

[paste bullet]

Prompt 48

Create a 30-day plan for a manual tester moving into Playwright automation
with 2 hours per day. Week 1: fundamentals. Week 2: first framework. Week 3:
CI + reporting. Week 4: portfolio project. Include daily deliverables.

Prompt 49

Draft a STAR answer for: 'Tell me about a time you found a critical bug
before release.' Situation, Task, Action, Result. Under 200 words. Honest and
specific — no dramatization.

Prompt 50

Write a salary-negotiation script for a QA Automation Engineer with 5 years
of experience receiving an offer that is 15% below market. Include the
research statement, the ask, a counter, and 2 benefits to negotiate if base
is fixed.

Practice these live with our AI Mock Interview and polish your resume via the ATS Resume Review.

Custom GPTs let you save instructions, upload knowledge files (PRD, API spec, coding standards), and reuse them without re-pasting context. These 10 prompts assume you've built a Custom GPT with your product docs loaded.

CG-1 — QA Standards GPT

Instructions: 'You are the SoftwareTestPilot QA Standards bot. Always use our
test-case column format (ID, Scenario, Preconditions, Steps, Data, Expected,
Priority, Type). Priorities are P0/P1/P2. Supported browsers: Chrome, Safari,
iOS Safari, Chrome Android. Never invent features not stated.'
Knowledge files: test-case-template.md, browser-matrix.md, DoD.md

CG-2 — Product PRD GPT

Instructions: 'You know Checkout v3.2. Answer only from the uploaded PRD.
When asked for test cases, use the QA Standards format. When a rule is
ambiguous, list it as an Open Question rather than assuming.'
Knowledge: checkout-prd-v3.2.pdf, api-spec-openapi.yaml, past-incidents.md

CG-3 — Automation Framework GPT

Instructions: 'You know our Playwright framework. Use BasePage, LoggedInFixture,
and our custom expect matchers. Never use waitForTimeout. Use role-based
locators. Return complete .spec.ts files.'
Knowledge: framework-readme.md, base-page.ts, fixtures.ts

CG-4 — Bug Triage GPT

Prompt: 'Here are 20 new bugs from last night [paste]. Cluster by likely
root cause, suggest owner team from CODEOWNERS knowledge, and rank by
blast-radius using our severity matrix.'

CG-5 — Release Notes GPT

Prompt: 'Given these merged PRs [paste titles + labels], draft customer-facing
release notes grouped by New Features, Improvements, Bug Fixes, Known Issues.
Use our tone guide (concise, benefit-first, no internal codenames).'

CG-6 — API Contract Guard GPT

Prompt: 'Diff this new OpenAPI spec against v3.2 (uploaded). List breaking
changes, deprecations, and required consumer updates for web/iOS/Android.'

CG-7 — Test Data Factory GPT

Prompt: 'Generate 100 realistic Indian users (name, email, phone, city,
GSTIN) with 60/30/10 split across free/pro/enterprise. Output as JSON matching
our User schema (uploaded).'

CG-8 — Interview Coach GPT

Instructions: 'You are a senior SDET interviewer at a FAANG-tier company.
Always ask one question at a time, wait for the answer, then score 1-5 on
correctness/structure/depth and give one probing follow-up.'
Knowledge: our-question-bank.md, rubric.md

CG-9 — Flaky Test Detective GPT

Prompt: 'Analyze this Playwright test-run history JSON. Identify tests with
failure rate > 5% over last 30 runs. For each, hypothesize root cause and
recommend fix (wait strategy, selector, cleanup, isolation).'

CG-10 — Release Risk GPT

Prompt: 'Given this sprint: [PRs merged], [tests added], [bugs open], [areas
changed], produce a go/no-go risk memo. Format: TL;DR verdict, top 3 risks,
mitigations, deferred items, sign-off owners.'

BEFORE PASTE — replace:
  Emails            -> user1@example.com
  Phone numbers     -> +91-9999999999
  Customer names    -> Customer A, Customer B
  Order IDs         -> ORD-0001
  Card numbers      -> 4111 1111 1111 1111 (test card)
  Tokens / API keys -> 
  Internal URLs     -> https://staging.example.com
  Employee IDs      -> EMP-001
  IP addresses      -> 10.0.0.1
  Screenshots       -> blur names, emails, IDs, revenue figures

Prefer ChatGPT Enterprise, Copilot for Business, or Gemini for Workspace — those do not train on your prompts. When in doubt, ask infosec before pasting.

1. Coverage — positive, negative, boundary, error paths all present?
2. Specificity — actual field names, error copy, business rules?
3. Data realism — plausible emails, PIN codes, valid card lengths?
4. Executability — can a junior tester run it without asking?
5. Traceability — maps to a requirement, story, or risk?
6. Safety — no destructive ops on prod-like envs?

Score each 1-5. Anything under 4 goes back for another prompt round — do not manually patch.

Is every case relevant to the actual feature? Are expected results clear? Is the test data realistic? Are high-risk areas covered? Can another tester execute the steps without asking basic questions? If yes, ship it.

ChatGPT can make software testers faster, but it will not replace observation, product understanding, or honest communication. The best QA engineers in 2026 use AI for speed and still bring human judgment for risk. Continue with our Claude AI test case generation guide, GitHub Copilot for QA, 15 best AI testing tools, and AI-powered bug detection tools.

Ground your AI-in-QA argument with original 2026 SoftwareTestPilot data:

• Enterprise automation tool benchmarks (2026) — where AI-augmented suites cut flake.
• Test pyramid benchmarks (2026) — real 62/26/12 layer ratios AI copilots should respect.
• Shift-left benchmarks (2026) — DORA-style ROI of AI-generated test coverage.
• Playwright market benchmarks — the runtime most AI test generators target.
• Practise the follow-up interview questions with the AI Mock Interview.