EasyMCP

Overview

Learning-prj is an MCP exploration workspace containing two complementary efforts: EasyMCP — a set of lightweight apps (daily info dashboard, personal notes, task reminders, local file handler) each built on a shared mcplearn server package — and MCP_all_App (mcp-core), a more complete modular FastAPI framework with SQLAlchemy, Alembic migrations, APScheduler, WebSocket support, and multi-platform adapters (Telegram, Twitter). Together they chart a progression from simple tool-based MCP servers toward a production-grade agentic backend.

Key Features

• Shared mcplearn base package providing BaseTool registration and MCP server scaffolding
• DailyTaskReminder: add, complete, list, and delete tasks with reminder checks via MCP tools
• LocalFileHandler: AI-assisted file naming (STT + LLM), create/move/rename tools, operation history log
• mcp-core framework: FastAPI transport layer, SQLAlchemy models, APScheduler jobs, Telegram and Twitter adapters
• Docker Compose deployment for mcp-core with PostgreSQL and Alembic migration workflow

Technical Decisions

The workspace uses a custom mcplearn package (distributed as a .whl alongside each app) to keep the MCP server boilerplate out of individual app code — each app only defines its tools and registers them. The mcp-core framework then extends this pattern toward a full production backend: replacing SQLite with PostgreSQL, adding WebSocket channels, and wiring multi-platform adapters behind a common dispatcher so new platforms can be added without changing the core.