Playwrightgen MCP Server

PlaywrightGen

AI-powered Playwright test generation platform for developers, automation engineers, and SDETs.

PlaywrightGen helps teams generate production-ready Playwright tests from prompts, components, HTML snippets, API descriptions, and page URLs.

Features

• Prompt → Playwright test generation
• Component → Playwright test generation
• Component → unit test generation
• HTML / JSX → Playwright test generation
• API test generation
• URL analysis for more realistic browser flows
• Multiple output styles: Fast, Clean, Production
• Free / Pro product structure
• Copy code and download output
• Built with Next.js + OpenAI

Example Use Cases

Developers

• Generate tests faster while building modern web applications

Automation Engineers

• Turn manual test ideas into automated Playwright coverage faster

SDETs

• Create stable locators, reusable flows, and production-ready test suites

Tech Stack

• Next.js
• TypeScript
• OpenAI API
• TailwindCSS
• Playwright
• Cheerio

Getting Started

Install dependencies:

npm install

Phase 1A workspace foundation

Checkpoint 2 adds the PostgreSQL schema and initial Prisma migration for the tenant-safe workspace. Checkpoint 4 synchronizes Clerk users, organizations, and organization memberships into those existing PostgreSQL models. Checkpoint 5 adds the server-only tenant authorization boundary, Checkpoint 6 adds tenant-scoped project and project-membership services with transactional Activity, and Checkpoint 7 adds the real database-backed workspace project list, creation, and overview experience.

The first V1 slice adds project-scoped Requirements, immutable RequirementVersion snapshots, explicit draft/review/approve/archive workflow, and real permission-aware Requirements routes inside each project.

AI Requirement Review stores advisory, evidence-linked suggestions against one immutable version. Suggestions can be accepted or dismissed but never mutate Requirement content. OPENAI_REQUIREMENT_REVIEW_MODEL optionally overrides the default review model; OPENAI_API_KEY remains server-only.

V1.3 adds versioned Test Cases with objective, preconditions, structured steps and expected results, review/approval, and direct Requirement traceability. Automation status is separate so future Playwright browser, API, and integration engines can create reviewable artifacts from approved versions.

V1.4 adds Test Runs pinned to an approved TestCaseVersion. Manual, Playwright Browser, and API modes store environment/browser metadata and append immutable attempts with overall/per-step results, duration, failure details, and evidence links. Retrying creates another attempt instead of overwriting history.

V1.5 adds advisory AI Failure Intelligence for failed and blocked attempts. It uses the OpenAI Responses API with Structured Outputs, validates every cited quote against immutable stored evidence, records model/prompt/schema/token metadata, and requires a Lead or Owner/Admin to confirm or dismiss findings. OPENAI_FAILURE_ANALYSIS_MODEL optionally overrides the default analysis model.

Phase 1A uses these environment variable names:

• DATABASE_URL: the PostgreSQL connection used by the application at runtime and by local development migration work when no direct connection is configured.
• DIRECT_URL: the optional direct PostgreSQL connection preferred by Prisma migration commands when a pooled runtime connection is unsuitable.
• TEST_DATABASE_URL: a physically separate PostgreSQL database used only by integration tests. It must identify a dedicated test database and never falls back to either development variable.
• NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY
• CLERK_SECRET_KEY
• CLERK_WEBHOOK_SIGNING_SECRET

V1.8 GitHub setup additionally requires separate development/Preview values:

• GITHUB_APP_ID, GITHUB_APP_PRIVATE_KEY
• GITHUB_APP_SLUG, GITHUB_APP_CLIENT_ID, GITHUB_APP_CLIENT_SECRET
• GITHUB_WEBHOOK_SECRET
• GITHUB_SETUP_STATE_SECRET (at least 32 characters and distinct from every other secret)
• NEXT_PUBLIC_APP_URL (the exact application origin used by the GitHub callback)

The GitHub App is read-only during repository import: metadata and contents read, with only installation and installation_repositories webhook events. User and installation access tokens are transient and must never be logged or stored.

Configure a Clerk Dashboard webhook endpoint at /api/webhooks/clerk and set CLERK_WEBHOOK_SIGNING_SECRET locally and in the deployment environment. Subscribe it to:

• user.created, user.updated, user.deleted
• organization.created, organization.updated, organization.deleted
• organizationMembership.created, organizationMembership.updated, organizationMembership.deleted

Local webhook delivery requires a secure tunnel to the local application. Do not place webhook secret values in documentation or source control.

Reconcile one Clerk organization without writing by default:

npm run clerk:reconcile -- --organization-id org_example
# or
npm run clerk:reconcile -- --slug workspace-slug

After reviewing the dry-run counts, explicitly add --apply to write the scoped reconciliation. The command never sweeps all organizations by default.

Database commands:

npm run prisma:format
npm run prisma:validate
npm run prisma:generate
npm run prisma:migrate:dev
npm run test:database

Never commit credentials, secrets, .env, or .env.local. Never run development migration or database reset commands against production from a developer laptop.