MCP Servers
Properties
created
28.03.2025, 14:43
modified
11.11.2025, 14:52
published
Empty
sources
MCP Server,
OpenAI SDK MCP,
MCP Spec Server Features
topics
MCP
authors
Empty
ai-assisted
No
The MCP Server is a foundational component in the Model Context Protocol (MCP) architecture that provides tools, resources, and capabilities to clients. It implements the server-side of the protocol, responsible for:
- Exposing tools that clients can discover and execute
- Managing resources with URI-based access patterns
- Providing prompt templates and handling prompt requests
- Supporting capability negotiation with clients
- Implementing server-side protocol operations
- Managing concurrent client connections
- Providing structured logging and notifications
# Custom Servers
People are writing custom servers for the MCP protocol to support additional data sources or software.
Sites for listing MCP servers:
- MCP.so - wbesite
- disco.dev
- Glama
Examples of Custom servers: - Blender MCP
- AWS S3 MCP
- The AWS MCP Server is now generally available
- 2026-05-06 — Managed remote MCP server giving agents authenticated access to all AWS APIs via
call_aws,search_documentation,read_documentation, andrun_script(sandboxed Python)
- 2026-05-06 — Managed remote MCP server giving agents authenticated access to all AWS APIs via
- New in Claude Managed Agents self-hosted sandboxes and MCP tunnels
- 2026-05-19 — MCP tunnels (research preview) let Claude Managed Agents reach private MCP servers via a single outbound gateway without exposing them publicly; paired with self-hosted sandboxes for in-perimeter tool execution
# Server Types
OpenAI SDK MCP:
Currently, the MCP spec defines two kinds of servers, based on the transport mechanism they use:
- stdio servers run as a subprocess of your application. You can think of them as running “locally”.
mcp.run(transport="stdio")
- HTTP over SSE (Server-Sent Events) servers run remotely. You connect to them via a URL.
mcp.run(transport="sse")
# Server Features
MCP Spec Server Features:
MCP servers provide three primitives with which LMMs can interact:
| Primitive | Control | Description | Example |
|---|---|---|---|
| Prompts | User-controlled | Interactive templates invoked by user choice | Slash commands, menu options |
| Resources | Application-controlled | Contextual data attached and managed by the client | File contents, git history |
| Tools | Model-controlled | Functions exposed to the LLM to take actions | API POST requests, file writing |
# Patterns
- MCP Server Patterns
- 2026-07-05 — catalogs five recurring server patterns (Resource Gateway, Tool Orchestrator, Stateful Session Server, Proxy Aggregator, Domain-Specific Adapter) plus four anti-patterns