clouatre-labs/math-mcp-learning-server

Metadata & Documentation Accuracy Updates

This release fixes critical inaccuracies in educational documentation and updates project metadata to reflect current capabilities.

pyproject.toml Updates

Metadata Improvements:

• Updated description to emphasize production-ready workspace persistence and FastMCP Cloud support
• Added Python 3.13 classifier (now officially supported)
• Enhanced keywords: added "cloud", "deployment", "workspace", "persistence"
• Added Python version upper bound <3.14 for stability

FUTURE_IMPROVEMENTS.md Fixes

Critical Corrections:

• Fixed model references throughout: Claude Sonnet 4 → Claude Sonnet 4.5
• Replaced "official MCP Python SDK patterns" with FastMCP 2.0 best practices
• Corrected anti-patterns section: removed [dependency groups] (FastMCP 2.0 actually uses [project.optional-dependencies])
• Updated all installation commands to FastMCP 2.0 syntax

New Content:

• Added FastMCP Cloud deployment section (Phase 1.3 - COMPLETE)
  • Cloud URL: https://math-mcp-learning.fastmcp.app/mcp
  • Zero-config deployment capabilities
  • Persistent workspace across cloud sessions
• Fixed v0.2.0 release date precision (September 28, 2025)

Pattern Updates:

• Replaced uv run mcp dev/install with proper FastMCP 2.0 commands
• Updated graceful degradation pattern to reference optional-dependencies
• Fixed pyproject.toml example: fastmcp>=2.0.0 (not mcp>=1.14.1)

Why This Release?

Educational accuracy is critical. The previous documentation:

• Referenced wrong Claude model version (Sonnet 4 instead of 4.5)
• Confused MCP SDK patterns with FastMCP 2.0 (different frameworks)
• Listed FastMCP 2.0 standards as "anti-patterns"
• Missing production cloud deployment capability

These fixes ensure learners get accurate, current information about MCP development.

Installation

No changes to functionality - this is purely documentation and metadata:

# Install from PyPI (unchanged):
uvx math-mcp-learning-server

# With plotting support:
uvx math-mcp-learning-server

# Cloud deployment (unchanged):
# Visit https://math-mcp-learning.fastmcp.app/mcp

Testing

✅ All 53 tests pass (1.48s)
✅ No code changes - documentation and metadata only

Full Changelog: https://github.com/huguesclouatre/math-mcp-learning-server/compare/v0.6.3...v0.6.4