Skill

MCP (Model Context Protocol)

Open JSON-RPC protocol that connects AI agents to tools, resources and prompts — I have built servers, clients and adapters for it.

  • #ai
  • #protocols

The Model Context Protocol (MCP) is a JSON-RPC specification for connecting AI hosts (Claude Desktop, Cursor, custom agents) to external systems. A server exposes three capabilities — tools, resources and prompts — over stdio or Streamable HTTP, and a client negotiates and calls them.

How I use it

  • mcp-deep-dive — a protocol-first TypeScript reference with a custom server and custom client on the raw SDK, all three capabilities, both transports and wire-level JSON-RPC frame logging. No LangChain wrappers.
  • Akashi — the akashi_mcp package lets Dart agents consume MCP tools; exposing an Akashi agent as an MCP server is on the roadmap.
  • Stores Helper — an MCP-style CLI that exposes App Store and Play Store management as tools for coding agents.
  • Conversational AI Agents — an MCP server inside a TypeScript monorepo in my Independent AI & Full-Stack Engineer (Freelance) work.
  • Somnio Software — MCP is part of the stack in my current role.

Credentials and courses

Developing MCP-Powered Agentic AI Systems, part of the Agentic AI Engineering Specialization specialization.

Related