model-context-protocol

# Model Context Protocol (MCP) Implementation This repository contains implementations for the Model Context Protocol (MCP), including both client and server components for interacting with external tools and services. ## Repository Structure ### Root Level - `client.py` - Standalone MCP client implementation - `README.md` - This file ### MCP Client (`mcp-client/`) A Python-based MCP client that connects to MCP servers and processes queries using Claude AI. **Features:** - Connect to MCP servers (Python or Node.js) - Interactive chat loop for querying - Tool execution and result processing - Integration with Anthropic's Claude API **Key Components:** - `client.py` - Main client implementation with MCPClient class - `pyproject.toml` - Project dependencies and configuration ### MCP Iceberg Server (`mcp-iceberg-server/`) A specialized MCP server for working with Apache Iceberg tables through AWS Glue. **Features:** - Database and table management - Schema inspection and metadata retrieval - Data querying with filtering capabilities - Table creation with custom schemas - Data insertion and upsert operations - Snapshot history tracking **Key Components:** - `iceberg_mcp_server.py` - Main server implementation using FastMCP - `pyproject.toml` - Project dependencies and configuration **Available Tools:** - `list_databases()` - List available databases - `list_tables()` - List tables in a database - `get_table_schema()` - Get table schema details - `get_table_data()` - Query table data with limits - `query_table()` - Filter table data by column values - `get_table_snapshots()` - View table snapshot history - `create_table()` - Create new tables with custom schemas - `insert_data()` - Insert or upsert data into tables ## Usage ### MCP Client ```bash cd mcp-client python client.py <path_to_server_script> ``` ### MCP Iceberg Server ```bash cd mcp-iceberg-server python iceberg_mcp_server.py ``` ## Configuration The Iceberg server supports configuration through environment variables: - `AWS_DEFAULT_REGION` - AWS region (default: us-east-1) - `ICEBERG_CATALOG` - Catalog name (default: s3tablescatalog) - `ICEBERG_DATABASE` - Database name (default: pyiceberg) - `ICEBERG_TABLE_BUCKET` - S3 bucket for tables (default: claude-table-bucket) - `LF_ROLE_ARN` - Lake Formation role ARN for permissions