Model Context Protocol

• MCP Docs:

MCP is an open protocol that standardizes how applications provide context to LLMs. Think of MCP like a USB-C port for AI applications. Just as USB-C provides a standardized way to connect your devices to various peripherals and accessories, MCP provides a standardized way to connect AI models to different data sources and tools.

• MCP Architecture
• MCP Primitives
• MCP Servers

• The 2026-07-28 MCP Specification Release Candidate
  • 2026-05-21 — RC for the next spec: stateless protocol core, Extensions framework, Tasks, MCP Apps, OAuth/OIDC-aligned authorization, formal deprecation policy

• FastMCP
  • The fast, Pythonic way to build MCP servers and clients
  • Like FastAPI but for MCP
    • So basically go to option for any project
  • Many additional features such as Code Mode, Tool Search etc.
  • jlowin/fastmcp
    • 23.2k stars
  • Resources
    • Introducing FastMCP 3.0
• MCP - Interactive user interfaces
  • MCP Apps: Extending servers with interactive user interfaces
• Test & debug MCP servers
  • https://www.mcpjam.com/
  • MCPJam/inspector
    • 1.4k stars

• GitHub MCP Exploited
• OpenAI adds MCP support to Agents SDK | Hacker News
• Originally no support for authentication and therefore limited support for remote MCP servers

• Claude can now connect to your world:

Developers can also create their own Integrations in as little as 30 minutes using our documentation or solutions like Cloudflare that provide built-in OAuth authentication, transport handling, and integrated deployment.

Source: How to use Anthropic MCP Server with open LLMs
MCP is originally designed for Claude but can be used with other LLM by writing custom adapter for it.
Official Support:

• OpenAI - openai-agents-python/mcp/
• Microsoft - MCP in Azure AI Foundry
  Custom adapters implementations:
• MCP: Integrating Azure OpenAI
  • Github: mcp_aoai
• Github: philschmid/mcp-openai-gemini-llama-example
• Github: jalr4ever/Tiny-OAI-MCP-Agent
• Github: chrishayuk/mcp-cli