Show HN: Tsar-MCP – How to Create an MCPServer Aspect in C/C++
Tsar-MCP is a learning guide for creating an MCP (Model Context Protocol) server in C/C++ with no external dependencies. It covers I/O flow, JSON parsing (using a decade-old JSON parser), and request dispatch, and provides an asynchronous, enterprise-ready baseline for building custom MCP servers.
Background
- MCP (Model Context Protocol) is an open protocol developed by Anthropic that lets AI models (like Claude) interact with external tools, data sources, and services in a standardized way — think of it as a "USB-C port" for connecting LLMs to the outside world.
- An MCP Server is a program that exposes specific capabilities (e.g., file search, database queries, API calls) to an AI assistant across a JSON-RPC channel.
- This guide teaches how to build such a server from scratch in C/C++ — unusual because most MCP servers are written in higher-level languages like Python or TypeScript. Doing it in C means the author had to write his own JSON parser (a non-trivial task) rather than relying on standard libraries.
- TSAR (Time Series Anomaly Reasoning) is IBM's broader open-source project; this MCP server is one component of it. The "aspect" in the title refers to a modular plug-in system within TSAR.
- The Show HN post indicates this is a personal/open-source project, not an official IBM product release.