🦅 wpipe-mcp

Transform your AI Agents into expert WPipe Architects.

wpipe-mcp is a professional Model Context Protocol (MCP) server that bridges the gap between AI Agents (Claude, Gemini, OpenCode) and the wisrovi SUITE. It empowers agents to search, design, validate, and deploy high-performance pipelines following strict industry-standard patterns.

✨ Key Features

• 🔍 Expert Catalog Search: Query over 100+ production-ready steps from Official and Community registries.

• 🏗️ Strict Architecture Enforcement: Guides AI to output clean code using mandatory dto/, states/, and main.py folder structures, including the new internal microservice app/ layout.

• 🛡️ Project Architectural Validation: Verify that any existing code structure complies with the WPipe standards using validate_wpipe_project to check directories, step decorators, and context schemas.

• ⚙️ Automated Monolith Refactoring: Automatically split monolithic Python scripts into modular class-based WPipe steps (@step) and shared contexts with refactor_monolith_to_wpipe.

• 📊 Automatic Pipeline Documentation: Generate complete Mermaid DAG flowcharts and technical markdown specifications directly in the project's root README.md with document_wpipe_project.

• 🧪 Unit Test Generation: Instantly build boilerplate test files with extensive English descriptions using generate_wpipe_tests.

• 🔍 Static Context Data Flow Analysis: Statically verify that data fields passed and consumed across steps are valid with validate_context_flow.

• ⚙️ Pipeline Dry Run Simulation: Simulate pipeline runs to trace context states without invoking heavy models/APIs using dry_run_pipeline.

• ⚡ Performance & Monitoring Optimization: Statically inspect compliance scores and receive actionable recommendations using optimize_wpipe_pipeline.

• 🏆 WPipe Quality Rules Enforcement: Built-in rules for strict English, Google-style docstrings, type hinting (mypy), Pylint score >= 9.3, Dockerized pytest coverage (>80%), and Black/Isort formatting.

• 📘 Architect's Manual: Built-in expertise for Monolith-to-Pipeline refactoring, performance tracking, and resilient state management.

• 💻 Unified CLI: Manage your MCP service with simple commands: run, start, stop, and config.

• 🔒 Privacy First: 100% local execution via stdio or SSE.

🛠️ Key Technologies & Libraries

This MCP server relies on the following key tools and libraries:

• Model Context Protocol (MCP): Server protocol implementation for agent interaction.
• FastMCP (mcp-sdk-python): Framework to declare tools and resources cleanly.
• Pydantic: Structural validation, schema checks, and context enforcement.
• Pytest & Pytest-Cov: Automated unit testing and coverage report analysis.
• Ruff: Modern linting and styling consistency.

🚀 Quick Start

1. Installation

Clone the repository and run the automated installer, or install via pip:

pip install -e .

2. Integration

Get your agent-specific configuration block and installation commands by running:

wpipe-mcp config

The CLI will dynamically detect your Python environment and provide exact copy-paste commands for Gemini CLI (e.g. gemini mcp add ...) and JSON blocks for Claude Desktop.

🧪 Running Tests

Unit tests are written with pytest. You can run them in three ways:

1. Locally (with development dependencies installed)

To execute tests locally with python path environment set up:

PYTHONPATH=src pytest tests/ -v

2. Inside a Docker Container (Recommended)

To run tests in a containerized environment to isolate dependencies:

./run_tests_docker.sh

3. Calculate Code Coverage

To run tests and get a detailed statement of code coverage:

./run_coverage.sh

This generates an HTML report in htmlcov/index.html.

📂 Project Structure

• src/wpipe_mcp/: Core server logic and tools.
• src/wpipe_mcp/catalog.py: Real-time GitHub catalog synchronization.
• src/wpipe_mcp/templates.py: Professional boilerplate definitions.
• examples/: Sample implementations and use cases.
• tests/: Automated unit tests verifying blueprints, scaffolding, and validation.

📄 License

MIT License - Crafted with ❤️ by William Rodriguez (wisrovi).