clouatre-labs/math-mcp-learning-server

🚀 What's New

FastMCP Cloud Deployment Support

This release adds full support for deploying to FastMCP Cloud with automatic configuration and production-ready defaults.

Live Demo: https://math-mcp-learning.fastmcp.app/mcp

New Configuration

• Added fastmcp.json for seamless cloud deployment
• HTTP transport configuration with INFO-level logging
• Automatic dependency management (fastmcp, pydantic, matplotlib, numpy)

Documentation

• Comprehensive cloud deployment guide in README
• Step-by-step deployment instructions
• Testing examples for FastMCP Client and Claude Desktop
• Cloud storage behavior documentation (ephemeral vs persistent)

🐛 Bug Fixes

Cloud Deployment Import Error Fix

Fixed critical import errors that prevented workspace persistence features from working in cloud deployments.

Problem:

ModuleNotFoundError: attempted relative import with no known parent package

Solution:
Converted all relative imports to absolute imports throughout the persistence module:

# Before (broken in cloud)
from .models import WorkspaceVariable

# After (works everywhere)
from math_mcp.persistence.models import WorkspaceVariable

Files Fixed:

• src/math_mcp/persistence/__init__.py (3 imports)
• src/math_mcp/persistence/workspace.py (2 imports)
• src/math_mcp/server.py (4 imports)

Impact:

• ✅ save_calculation now works in cloud
• ✅ load_variable now works in cloud
• ✅ math://workspace resource now accessible in cloud
• ✅ Complete feature parity between local and cloud deployment

📊 Testing

All tests passing:

• ✅ 38 unit tests passed
• ✅ Cloud deployment tested and verified
• ✅ Workspace persistence functional in cloud (ephemeral)
• ✅ Backward compatible with local development

🌐 Deployment

FastMCP Cloud:

• Automatic deployment from main branch
• Server URL: https://math-mcp-learning.fastmcp.app/mcp
• Public access (no authentication required)
• Discoverable in FastMCP marketplace

Local Installation:

# PyPI
uvx math-mcp-learning-server
uvx math-mcp-learning-server

# Claude Code
claude mcp add mathmcp uvx math-mcp-learning-server[plotting]@latest

# Cloud version
claude mcp add math-mcp-cloud https://math-mcp-learning.fastmcp.app/mcp --transport http

📝 What's Changed

Full Changelog: https://github.com/huguesclouatre/math-mcp-learning-server/compare/v0.6.0...v0.6.1

Key PRs:

• #26: Add FastMCP Cloud deployment configuration
• #27: Fix relative imports for cloud compatibility
• #25: Add CODE_OF_CONDUCT.md and remove CLAUDE.md from repo

🔄 Breaking Changes

None. This is a backward-compatible patch release.

🎓 Educational Impact

This release demonstrates:

• FastMCP 2.0 cloud deployment best practices
• Python import patterns for cloud-native applications
• Transport-agnostic MCP server design
• Professional open source project management (Code of Conduct)

📚 Documentation

• Cloud Deployment Guide
• Contributing Guidelines
• Code of Conduct

Contributors: Hugues Clouâtre