The Rust MCP SDK, developed by Paiml, is a Rust-based SDK distributed as the pmcp crate. It serves as a production-grade solution for connecting to the Model Context Protocol (MCP) ecosystem. This SDK provides both client and server implementations, enabling communication between AI models and their runtime environments. It is designed for developers building high-performance, multi-agent systems and multi-modal AI processing pipelines.

What does Rust MCP SDK connect to?

The Rust MCP SDK, distributed as the pmcp crate, establishes connections within the Model Context Protocol (MCP) ecosystem. It offers both client and server implementations, positioning itself as a production-grade SDK for MCP servers and clients. This design enables essential communication pathways between AI models and their respective runtime environments. Its primary category is supporting these critical interactions.

Core features and tools

The SDK exposes primitives mirroring the MCP specification, including tools for callable functions, prompts for parameterized templates, and resources for URI-addressable content. Developers define specific callable tools using macros like #[mcp_tool], #[mcp_prompt], and #[mcp_server]. Clients discover tools via list_tools and invoke them via call_tool. It supports five transport protocols: STDIO, HTTP/1.1, HTTP/2, WebSocket, and SSE. WASM is also a transport option. It works with AI clients such as ChatGPT and Claude Desktop, and any TypeScript client.

Who is it for?

The Rust MCP SDK targets backend engineers and AI developers focused on building production MCP servers. It is particularly suited for those creating high-performance, multi-agent systems and multi-modal AI processing pipelines. The SDK supports advanced workflows requiring distributed AI agent networks and complex data processing. Skip this if your project does not involve building MCP servers or interacting with AI models via this protocol.

How to get started

Access is granted through a standard OAuth authorization flow, supporting full authentication context pass-through. The MCP server verifies tokens from other systems or integrates with external identity providers, including remote OAuth providers using Dynamic Client Registration.

Use cases

• Building distributed AI agent networks with bidirectional communication, circuit breakers, and automatic failover.
• Creating complex data processing workflows for text, images, audio, and structured data, including streaming and batch operations.
• Building SQL and OpenAPI/HTTP MCP servers from configuration files, or serving governed Excel workbooks from compiled bundles.

The Rust MCP SDK provides a foundational framework for developing production-grade MCP clients and servers. It offers essential primitives and supports various transport protocols, making it adaptable for diverse AI communication needs. Its authentication model supports OAuth for secure interactions. This SDK is ideal for engineers building advanced AI agent systems and data pipelines.