A comprehensive workspace dedicated to the Model Context Protocol (MCP), featuring a suite of servers, clients, and automated workflows designed to bridge the gap between Large Language Models (LLMs) and real-world data environments.
This repository serves as a centralized hub for custom-built MCP servers, UI integrations, and orchestration layers that enable sophisticated AI agents to interact with local files, APIs, and databases.
Custom-built servers providing specialized tools for LLM consumption:
- Data Extractors: PDF parsing and structured data extraction (using
pdf-parseandzod). - Reporting APIs: Custom interfaces for real-time data fetching and analysis.
- System Utilities: Tooling for local file system management and command execution.
Interfaces and wrappers that bring MCP capabilities to the end-user:
- Desktop Apps: Cross-platform UI developed with Avalonia UI, integrating Node.js logic for a seamless desktop experience.
- CLI Tools: Lightweight terminal-based clients for rapid testing and server interaction.
Advanced orchestration combining the best of the AI ecosystem:
- LangChain Integration: Chains and agents that leverage MCP tools for complex, multi-step reasoning.
- Vector Search: Implementation of RAG (Retrieval-Augmented Generation) patterns using local or hosted vector databases.
- Database Management: Seamless connectivity between MCP servers and SQL/NoSQL databases for persistent state management.
- Core Protocol: Model Context Protocol (MCP)
- Runtimes: Node.js, .NET (Avalonia)
- Languages: TypeScript, C#
- AI Frameworks: LangChain, Ollama (Local SLMs/LLMs)
- Data Handling: Vector DBs, Zod (Schema Validation)
├── apps/ # Finished applications (Avalonia, etc.)
├── servers/ # MCP Server implementations (Node.js/TypeScript)
├── workflows/ # LangChain scripts and automation logic
├── packages/ # Shared utilities and types
└── testing/ # Test suites for protocol compliance