UniAuto MCP Server is a unified automation platform that combines web and desktop test automation with self-healing capabilities and AI integration. It exposes an MCP-compliant API so AI assistants can orchestrate tests, generate and run automation scripts, manage test cases, and perform advanced checks like visual comparisons and accessibility tests. With built-in support for Playwright-based web automation, desktop automation capabilities, and AI-driven test generation and maintenance, UniAuto empowers both engineers and non-technical stakeholders to define and execute resilient tests via natural-language prompts or structured API calls.

To use the server, install dependencies, configure environment variables (including AI integration keys), and start the server. The MCP endpoints let you invoke automation commands, manage test cases, generate tests and full suites, scaffold projects, and leverage the MCP manifest to understand capabilities. When connected to an AI assistant (e.g., Claude via Smithery.ai), the AI can issue MCP commands to execute flows, generate selectors, perform visual comparisons, and produce test code in multiple frameworks (Playwright, Cypress, Jest, etc.). The integration with Smithery.ai enables seamless connection between the UniAuto MCP server and AI orchestration platforms for coordinated test automation tasks.

Prerequisites:

• Node.js and npm installed
• Git installed
• Optional: Playwright browsers if running web automation locally
• Access to Claude API (if using AI integration) and corresponding API keys

Installation steps:

1. Clone the repository: git clone https://github.com/yourusername/uniauto-mcp-server.git cd uniauto-mcp-server

2. Install dependencies: npm install

3. Install Playwright browsers (for web automation): npx playwright install

4. Create and customize environment file (example): cp .env.example .env

   Edit .env and populate required values (see Configuration section below)

5. Start the server: npm run dev # for development mode

   or

   npm start # for production mode

Environment configuration (example):

• PORT=3000
• NODE_ENV=development
• MONGODB_URI=mongodb://localhost:27017/uniauto
• CLAUDE_API_KEY=your_claude_api_key
• CLAUDE_MODEL=claude-3-7-sonnet-20240229

Tips and common considerations:

• Ensure your CLAUDE_API_KEY is valid and the model name matches your plan; update CLAUDE_MODEL as needed.
• If running without MongoDB, adjust MONGODB_URI or disable database usage as appropriate.
• When using MCP with an AI assistant, ensure the assistant is configured to discover and invoke the MCP manifest exposed by /api/mcp/manifest.
• For production deployments, consider mounting an environment file securely and configuring proper logging and monitoring (e.g., PM2, systemd).
• The API includes endpoints for health checks, automation commands, test management, AI processing, test generation, MCP invocations, and manifest retrieval. Use /api/mcp/manifest to understand capabilities exposed to MCP clients.
• If visual comparison or accessibility tests fail due to resource paths, verify the baseline assets and ensure file permissions are correct.
• Keep dependencies up to date to benefit from improvements in Playwright, MCP protocol support, and AI integrations.