🔗 Browse it live: qa-skill-library.vercel.app
- What it is — 28 reusable AI skills covering the QA lifecycle: requirement analysis, test planning, test design, test case development, execution, defect management, RCA, and test closure, plus a dedicated Playwright automation pack.
- Why it matters — you shouldn't have to re-explain your QA process to an AI assistant every time you open a new chat. Each skill writes the workflow, the expected output, and the guardrails down once, in a reusable
SKILL.mdfile the whole team shares. - Value delivered — it turns the assistant into a repeatable QA teammate rather than a chatbot you brief from scratch, producing structured drafts for test plans, test cases, defect triage, RCA, coverage analysis and closure — with human review before anything is final.
New to this? A skill is just a text file that tells your AI assistant how to do one QA job — like the onboarding note you'd hand a new tester, except the assistant re-reads it before every run. Nothing to install: point your assistant at the SKILL.md path and ask.
A library of reusable AI assistant skills that cover the Software Testing Life Cycle (STLC) end to end — from judging whether a requirement is ready to test, through planning, design, test case authoring, defect management, and test closure. Each skill is a single SKILL.md file: a self-contained set of instructions that tells the AI assistant exactly how to perform one QA activity, consistently, every time.
- QA Engineers — run any skill end-to-end, from readiness checks through closure reports.
- SDETs / Automation Engineers — take an approved scenario or test case straight into the Playwright Automation Pack.
- Test Leads — triage backlogs, run RCAs, and produce closure reports for sign-off.
- Business Analysts — pressure-test requirements for gaps and ambiguity before they reach development.
- Product Owners — use requirement-readiness and test-planning output to validate scope before committing to a sprint.
This library provides reusable AI assistant skills across the entire Software Testing Life Cycle — requirement analysis through closure, not just test design or automation:
Requirement Analysis
│
▼
Test Planning
│
▼
Test Design
│
▼
Test Case Development
│
┌────┴─────┐
▼ ▼
Test Data Playwright Automation
│ │
└────┬─────┘
▼
Test Execution
│
▼
Defect Management
│
▼
Root Cause Analysis
│
▼
Test Coverage Analysis
│
▼
Test Closure
Every box above is backed by one or more reusable AI assistant skills — see Skill Catalog below.
Skills are independent: any box can be run on its own. Previous outputs are reused when available (e.g., a Test Plan reuses an existing Readiness Assessment) — but no hand-off is required to get value from a single skill.
Manually re-explaining "how we do requirement analysis" or "how we triage bugs" to an AI assistant every time is slow and inconsistent — one session's format won't match the next, and important guardrails (never invent an acceptance criterion, never auto-close a bug, always stop for human sign-off) can quietly get skipped. This library fixes that by encoding each QA activity once, as a skill, so that:
- The same request ("is this ticket ready to test?") produces the same structured, review-ready output every time, from anyone on the team.
- Every skill is scoped to one responsibility (e.g., Test Scenario Designer decides what to test, never how — that's the Playwright Automation Pack's job) so outputs stay predictable and skills can be chained safely.
- Every skill treats its output as a draft for a human to confirm, never a final decision — nothing gets written back to Jira, closed, merged, or shipped by the skill itself.
A skill is a Markdown file (SKILL.md) with a short YAML header (name, description, version) followed by structured instructions: what inputs it accepts, what workflow it follows step by step, what its output must contain, and what it must never do (its guardrails). A compatible AI assistant reads the skill file and follows it in place of general-purpose reasoning — the same way a new teammate would follow a documented procedure instead of improvising.
You invoke a skill by pointing your AI assistant at its SKILL.md path (or describing your task the way its description trigger expects — see the Example Prompt column in the Skill Catalog below) and telling it what to produce.
| Benefit | What It Means in Practice |
|---|---|
| Consistency | Every readiness assessment, test plan, or bug report follows the same sections and terminology, regardless of who asked for it. |
| Traceability | Every finding, scenario, or test case must trace back to a stated requirement, AC, or piece of evidence — nothing is invented to fill a gap. |
| Human-in-the-loop by design | Every skill ends in a mandatory Human Review Gate (Facts / Observations / Assumptions / Unknowns) — the assistant proposes, a person confirms. |
| Independent & composable | Every skill works standalone — you can jump straight to Test Case & RTM Generator without ever running the earlier skills. When an upstream output is available, downstream skills reuse it instead of re-deriving it — but reuse is optional, never required. |
| Faster QA cycles | Turns a Jira ticket ID into a structured, review-ready draft (readiness score, test plan, scenarios, RTM, bug report, RCA, closure report) in minutes instead of a manual write-up. |
| Confidence is honest, not inflated | Every skill states a Confidence Level (High/Medium/Low) tied to how complete the evidence actually was — never to how long the report is. |
- An AI coding assistant that supports skill files (e.g. Claude Code, GitHub Copilot, or another compatible agent/CLI) — the environment that reads and executes these
SKILL.mdfiles. - This skill library available locally — cloned or present in your workspace (e.g. this
SKILL_CATALOG). - Optional — Jira or Azure DevOps access — only needed if you want a skill to retrieve a requirement or defect directly by ID instead of pasting text.
- Optional — an MCP integration (e.g. Atlassian/Rovo) configured in your AI assistant — enables direct Jira/ADO retrieval above, and lets Bug Reporter automatically check Jira for an existing/duplicate defect before you file a new one.
- Optional — a Playwright project — only needed if you intend to run the drafts produced by the Playwright Automation Pack.
Skills are grouped into folders numbered by STLC phase, alongside a separate, standalone Playwright automation pack:
SKILL_CATALOG/
├── STLC_Skills/
│ ├── 01_requirement-analysis/
│ ├── 02_test-planning/
│ ├── 03_test-design/
│ ├── 04_test-case-development/
│ ├── 05_test-execution/
│ ├── 06_defect-management/
│ └── 07_test-closure/
└── Automation_Playwright/ ← standalone Playwright automation pack (15 skills;
not part of the numbered STLC phases above)
| # | STLC Phase | Skill | What It Does | When to Use It | Example Prompt |
|---|---|---|---|---|---|
| 1 | Requirement Analysis | Requirement Readiness Analyzer | Scores a requirement's readiness for QA — gaps, ambiguities, risks, clarifying questions. | Before test planning starts, or whenever a story looks thin. | "For ABC-1165, using STLC_Skills/01_requirement-analysis/requirement-readiness-analyzer/SKILL.md, assess readiness and create req_report.md." |
| 2 | Test Planning | Test Plan Generator | Turns a requirement into a scope, strategy, risk, and entry/exit-criteria plan. | Once a requirement is at least conditionally ready. | "For ABC-1165, using STLC_Skills/02_test-planning/test-plan-generator/SKILL.md, create a test plan in testplan.md." |
| 3 | Test Design | Test Scenario Designer | Derives a categorized, risk-tagged, AC-traced scenario matrix (WHAT to test). | After the test plan, before anyone writes step-by-step cases. | "For ABC-1165, using STLC_Skills/03_test-design/test-scenario-designer/SKILL.md, create test scenarios in testscenario.md." |
| 4 | Test Design | API Test Designer | Maps contract-level API coverage (schema, auth, negative, boundary, error, idempotency). | When testing an endpoint or OpenAPI/contract change. | "Using STLC_Skills/03_test-design/api-test-designer/SKILL.md, design API test coverage for POST /orders/{id}/checkout." |
| 5 | Test Case Development | Test Case & RTM Generator | Generates manual test cases, BDD scenarios, an RTM, and an automation assessment. | Once scenarios/ACs are approved and need step-by-step cases. | "For ABC-1165, using STLC_Skills/04_test-case-development/test-case-writer/SKILL.md, generate manual test cases and an RTM." |
| 6 | Test Case Development | Test Data Generator | Generates positive/negative/boundary/synthetic/bulk test data — never real customer data. | Whenever a test case needs concrete input values. | "Using STLC_Skills/04_test-case-development/test-data-generator/SKILL.md, generate boundary and negative test data for the shipping-address fields in ABC-1165." |
| 7 | Test Execution | Regression Suite Selector | Picks a risk-ranked regression subset for a change (Must-run/Should-run/Skip), plus a safety net — no PR/repo access required. | Before a PR merge, or whenever a full-suite run is too slow/costly for the change. | "We changed how checkout validates the promo code (no PR link, just describing it) — using STLC_Skills/05_test-execution/regression-suite-selector/SKILL.md, tell me what to re-run." |
| 8 | Test Execution | Test Execution Tracker | Logs pass/fail/blocked/not-run per case with evidence, then rolls up completion % and pass rate. | While a test cycle is underway, or for a standup/status snapshot. | "Using STLC_Skills/05_test-execution/test-execution-tracker/SKILL.md, log today's results for the ABC 1.2 cycle and give me the rollup." |
| 9 | Defect Management | Bug Reporter | Turns an observed failure into a reproducible, evidence-based defect report, and — optionally, if Jira MCP is connected — checks Jira for an existing/duplicate defect first. | The moment a tester finds something broken. | "In the ABC app I saw a blank screen after submitting the checkout form — using STLC_Skills/06_defect-management/bug-reporter/SKILL.md, create bug-report.md. And if Jira is connected, check for similar existing defects before recommending whether to create a new bug or update an existing one." |
| 10 | Defect Management | Bug Triage Assistant | Clusters likely duplicates, proposes severity/priority/routing across a defect batch. | Before a triage meeting, on a backlog or JQL export. | "Triage the open ABC bugs from the last 15 days using STLC_Skills/06_defect-management/bug-triage-assistant/SKILL.md; create triage.md." |
| 11 | Defect Management | RCA Analyzer | Runs 5-Whys + fishbone to separate root cause from symptom, proposes CAPA. | When a defect needs "why did this happen" beyond the symptom. | "Do an RCA of ABC-2415 using STLC_Skills/06_defect-management/rca-analyzer/SKILL.md; create rca.md." |
| 12 | Test Closure | Test Closure Reporter | Rolls execution/defect/coverage metrics into an advisory Go/No-Go closure report. | At the end of a test cycle, before sign-off. | "Using STLC_Skills/07_test-closure/test-closure-reporter/SKILL.md, write the closure report for the ABC 1.2 test cycle." |
| 13 | Test Closure | Test Coverage Analyzer | Finds the gap between what was required and what was actually tested/automated/executed. | Before closure, or whenever "are we actually covered?" comes up. | "Using STLC_Skills/07_test-closure/test-coverage-analyzer/SKILL.md, find coverage gaps for ABC-1165 against our RTM." |
Playwright Automation Pack (standalone, not a numbered STLC phase) — 15 skills covering test generation, fixtures, Page Objects, locator hardening, flaky-test diagnosis, trace analysis, API testing, network mocking, visual regression, mobile/responsive testing, multi-tab/popup handling, test-step reporting, CI configuration, suite-health reporting, and a suite auditor that routes findings to the right skill. See Automation_Playwright/ — start with PW Suite Auditor on an existing suite, or PW Test Generator once a scenario/case is approved and ready to automate.
| Skill | Example Prompt |
|---|---|
| PW Suite Auditor | "Using Automation_Playwright/pw-suite-auditor/SKILL.md, audit our existing Playwright suite and give me a prioritized migration plan." |
| PW Test Generator | "Turn scenario TS-1 for ABC-1165 into a Playwright test using Automation_Playwright/pw-test-generator/SKILL.md." |
| Skill | Input Required | Output Received |
|---|---|---|
| Requirement Readiness Analyzer | A requirement ID (Jira/ADO) or pasted user story / AC text | Readiness Score & Verdict, Findings by severity, Dependency Analysis, Clarifying Questions, STLC Progression Assessment |
| Test Plan Generator | A requirement ID or pasted AC (+ prior Readiness Assessment, if available) | Scope & Objectives, Gap Analysis, Test Strategy, high-level Scenario Summary, Risks & Assumptions, Entry/Exit Criteria, Planning Confidence |
| Test Scenario Designer | A requirement ID / ACs (+ prior Readiness/Plan output, if available) | Risk-tagged Scenario Matrix, Coverage Summary (per-AC status), Traceability Summary, Scenario Readiness Assessment |
| API Test Designer | An OpenAPI/Swagger spec, contract, or endpoint documentation | Endpoint Summary, Coverage Matrix, Assertion Summary, Risk-ranked coverage, Traceability Summary |
| Test Case & RTM Generator | A user story, AC, PRD/BRD, spec, wireframe, or screenshot | Manual Test Cases, BDD Scenarios, Automation Assessment, Requirement Traceability Matrix, Coverage Gap Analysis |
| Test Data Generator | Field/entity/schema definitions, ACs, test cases, or a DB/API schema | Positive/Negative/Boundary/Synthetic/Bulk/Relational data sets, Data Matrix, reusable data recipes |
| Regression Suite Selector | A diff/PR link, a changed-file list, or a plain-language change description | Impact Map, Must-run/Should-run/Skip-with-reason selection, Safety Net, Coverage Gap, Confidence Level |
| Test Execution Tracker | A case list/RTM, plus reported results (status, evidence, run owner) per case | Execution Log, Blockers, rollup Metrics (completion %, pass rate) with denominators, Confidence Level |
| Bug Reporter | An observed failure description, plus any screenshots/videos/logs/traces | Structured Bug Report — repro steps, expected/actual, severity/priority, evidence, suspected component, plus an optional Existing Defect Check (Jira MCP Status, Candidate Issues, Similarity Assessment, Recommendation) when Jira MCP is connected |
| Bug Triage Assistant | A defect list or backlog/JQL export | Duplicate Clusters, prioritized defects (severity/priority/routing), Backlog Health Summary |
| RCA Analyzer | A defect/incident description plus available evidence (logs, timeline, commits) | Problem Statement, Timeline, 5-Whys & Fishbone, Root Cause vs Contributing Factors, CAPA Plan |
| Test Closure Reporter | Execution results, defect data, coverage analysis, exit criteria | Computed Metrics, Coverage/Defect Assessment, Exit Criteria Evaluation, Residual Risk, advisory Go/No-Go |
| Test Coverage Analyzer | Requirements/ACs, an existing RTM, test cases, automation/execution reports | Traceability Matrix, Coverage Metrics, risk-ranked Gap Analysis, Orphan Tests |
See Automation_Playwright/ for the Playwright pack's own input/output reference — each of its 15 skills documents this in its own SKILL.md.
Skills are designed to hand off to each other, but no hand-off is mandatory — every skill can be run on its own. When a prior skill's output exists, the next skill reuses it rather than repeating the analysis:
Requirement Readiness Analyzer
│ (readiness assessment reused, not re-derived)
▼
Test Plan Generator
│ (plan scope/risks reused)
▼
Test Scenario Designer ──────┐ (or, for APIs) API Test Designer
│ (scenarios reused) │
▼ ▼
Test Case & RTM Generator ◄──┘
│
├──► Test Data Generator (data for the cases above)
└──► PW Test Generator (Automation_Playwright pack — automation draft for approved cases)
│
▼
Test Execution
Regression Suite Selector (pick what to run) ──► Test Execution Tracker (log the results)
Defect Management (runs independently, any time)
Bug Reporter (optional Jira duplicate check) ──► Bug Triage Assistant ──► RCA Analyzer
Test Closure (end of cycle)
Test Coverage Analyzer ──► Test Closure Reporter (consumes the Execution Tracker's rollup)
Each arrow is a reuse, not a re-analysis — e.g., the Test Plan Generator will consume an existing Requirement Readiness Assessment's findings if one exists, rather than re-scoring the same requirement from scratch; if one doesn't exist, it simply performs the analysis itself.
Regardless of which skill you use, all 13 STLC skills share the same non-negotiable rules:
- Never invents evidence — a missing acceptance criterion, log line, or field is reported as a gap or clarification question, never filled in.
- Never changes tracker state — no skill auto-closes, merges, reassigns, files, or marks a ticket as a duplicate on its own (including Bug Reporter's optional Jira duplicate check — it only recommends); you paste/file the output yourself once you've confirmed it.
- Always states confidence honestly — High/Medium/Low, tied to evidence completeness, not to how polished the write-up looks.
- Always ends in a Human Review Gate — Facts / Observations / Assumptions / Unknowns, requiring your confirmation before the output is treated as final.
| Component | Status |
|---|---|
| Core QA Skills (Requirement Analysis → Test Closure, 13 skills) | ✅ Complete |
| Playwright Automation Pack (Automation_Playwright/, 15 skills) | ✅ Complete |
Additional Playwright-specific documentation and examples will be added in future releases.
- Confirm you have the Prerequisites above in place.
- Pick the skill matching your task from the Skill Catalog.
- Give your AI assistant the skill's
SKILL.mdpath plus your input (a Jira ID, pasted text, a spec, or a description of what broke) — see the Example Prompt column in the Skill Catalog. - If your input is a Jira ticket, make sure the Atlassian/Rovo MCP connection is authenticated for this session (your AI assistant will prompt you the first time).
- Review the draft output against its Human Review Gate section before acting on it, filing it, or handing it to the next skill in the chain.
The banner image at the top of this README (QA_Skills_Catalog.png) is a static, at-a-glance overview of the whole suite.
👉 qa-skill-library.vercel.app — an interactive, searchable version of this catalog: all 28 skills, each with a copy-ready prompt. The source is website/index.html — a standalone, self-contained page with no build step, deployable as-is anywhere that serves static files.
Shivani Singh Senior QA Engineer | AI-Assisted Quality Engineering & Test Automation