Agent skills that turn AI coding assistants into senior-level SDETs — writing production-grade Playwright E2E tests, scaffolding projects, generating Page Object Models, and managing the full test lifecycle from handover through regression.
Install once, and your agent knows how to structure, write, and maintain end-to-end tests the way a senior SDET would. Skills follow the open Agent Skills standard and work with 40+ AI coding agents.
Fix the QA bottleneck AI created.
AI agents can generate code, open PRs, and accelerate delivery — but they still struggle to create and maintain reliable end-to-end tests.
This leaves QA as the bottleneck.
TestSkills bridges that gap by teaching agents how to:
- structure test suites correctly
- write stable Playwright tests
- manage the lifecycle from handover → regression
So speed doesn't come at the cost of confidence.
npx skills add AgentMantis/test-skillsOr install for a specific agent:
npx skills add AgentMantis/test-skills --agent claude-code
npx skills add AgentMantis/test-skills --agent cursor
npx skills add AgentMantis/test-skills --skill create-pom # single skill onlyThen ask your agent:
> Set up an e2e test suite for my project
> Create a page object for the login page
> Write a regression test for the vault feature
> Write a handover test for MANT-456
> Promote the MANT-456 test to regression
In Claude Code, skills also register as slash commands: /e2e-test-suite-init, /create-pom, /create-regression-test, /create-handover-test, /promote-handover-test.
Without TestSkills:
- inconsistent test structure
- brittle Playwright tests
- unclear promotion from handover → regression
- QA bottlenecks as AI accelerates delivery
With TestSkills:
- standardised test architecture
- repeatable Page Object Models
- lifecycle-driven testing (handover → regression)
- agents that behave like senior SDETs
| Skill | Type | Invoke | Description |
|---|---|---|---|
e2e-test-conventions |
Reference | auto-loaded | Core conventions — project structure, naming, selectors, auth, parallelism, environment config |
e2e-test-suite-init |
Task | /e2e-test-suite-init [name] |
Scaffolds a complete e2e/ directory with config, base classes, fixtures, and helpers |
create-pom |
Task | /create-pom [page] |
Creates a Page Object Model with lifecycle methods, locators, JSDoc, and comment banners |
create-regression-test |
Task | /create-regression-test [feature] |
Scaffolds a regression or smoke spec with test data JSON, fixtures, and beforeAll cleanup |
create-handover-test |
Task | /create-handover-test [TICKET] [desc] |
Creates a ticket-driven handover test with proper naming and promotion lifecycle |
promote-handover-test |
Task | /promote-handover-test [TICKET] |
Promotes a handover test to the regression or smoke suite |
Each skill is self-contained and can be installed independently with --skill <name>. The e2e-test-conventions reference skill is loaded automatically whenever any task skill runs.
Skills are designed to be used in sequence as your project grows:
e2e-test-conventions (auto-loaded — always active)
→ e2e-test-suite-init (run once per project)
→ create-pom (per page/view)
→ create-regression-test (per feature)
→ create-handover-test (per ticket)
→ promote-handover-test (when ticket is done)
These skills enforce best practices used by senior SDETs.
They prioritise:
- stability over cleverness
- readability over abstraction
- lifecycle-driven testing over ad-hoc scripts
Because AI-generated tests are only useful if they can be trusted.
Skills follow the open Agent Skills standard and work with 40+ AI coding agents via npx skills add, including Claude Code, Codex, Cursor, Antigravity, GitHub Copilot, Augment, Gemini CLI, Windsurf, Roo, and more.
Installation options
# Install all skills (auto-detects your agent)
npx skills add AgentMantis/test-skills
# Target a specific agent
npx skills add AgentMantis/test-skills --agent claude-code
# Install a single skill
npx skills add AgentMantis/test-skills --skill create-pom
# Install globally (available in all projects)
npx skills add AgentMantis/test-skills -g
# List available skills before installing
npx skills add AgentMantis/test-skills --listcp -r skills/* ~/.claude/skills/ # Claude Code (personal)
cp -r skills/* .claude/skills/ # Claude Code (project)Repository structure
test-skills/
├── skills/ # Source of truth (edit here)
│ ├── e2e-test-conventions/ # Reference skill (auto-loaded)
│ ├── e2e-test-suite-init/ # Scaffold a new e2e/ directory
│ ├── create-pom/ # Generate a Page Object Model
│ ├── create-regression-test/ # Generate a regression/smoke spec
│ ├── create-handover-test/ # Generate a ticket-driven spec
│ └── promote-handover-test/ # Promote handover → regression/smoke
├── docs/
│ └── CUSTOMIZATION.md # Adapt skills to your project
├── CONTRIBUTING.md # How to add or modify skills
├── CHANGELOG.md # Release history
├── LICENSE # MIT
└── README.md
Customization
These skills teach generic Playwright patterns that work with any UI framework. To adapt them to your project's specific base classes, fixtures, folder structure, or auth flow, see docs/CUSTOMIZATION.md.
TestSkills is part of the Agent Mantis ecosystem — a Testing-as-a-Service platform designed for AI-native development teams.
If you're using AI to ship faster, Agent Mantis helps you:
- validate changes continuously
- orchestrate testing workflows
- maintain confidence at scale
→ Learn more: https://agentmantis.com
The skills/ directory is the single source of truth for all skill content.
Only edit files inside skills/. Do not create or modify files outside this directory for skill content. Each skill lives in its own subdirectory with a SKILL.md and optional references/ folder — this is the standard format that npx skills add distributes to all agents.
We welcome contributions — new skills, bug fixes, and improvements to existing skills. See CONTRIBUTING.md for guidelines on the SKILL.md format, testing expectations, and how to submit a PR.
MIT — Copyright (c) 2026 Agent Mantis