koki7o

mcp-framework: Build Rust AI Agents with MCP Support

A Rust framework built on the official rmcp MCP SDK and Tokio for building AI agents, MCP clients and servers. It integrates Anthropic Claude and OpenAI, and includes a web inspector for testing tools.

QuietRustMITAI AgentsMCPFramework
View on GitHub Compare
Stars
25
Forks
1
Last commit
10 months ago
Contributors
1
Releases, 12 months
0

Our take

Quietno commits in the last 90 days and no release in a year

Health score

17/100

How it is scored
Activity
20
Community
4
Issues
21
Pull requests
21
  • Fewer contributors than 91% of the projects we track

This is a compact, MCP-focused Rust framework with useful agent and inspector examples, but its quiet maintenance and single-contributor history make it a cautious choice for production dependencies.

Good fit if

  • Your application is in Rust and you want MCP client, server and agent components built around rmcp and Tokio.
  • You need Claude or OpenAI integration and value the included web inspector for tool testing.
  • You can evaluate and maintain the framework yourself under its MIT license.

Look elsewhere if

  • You need a project with recent releases or an active contributor base for ongoing maintenance.
  • You require the roadmap features such as vector database integrations, agent orchestration, MCP resources or prompts.
  • Your deployment depends on a packaged Docker, Compose or Helm setup, which the project does not provide.
All health signals
Last commit2025-12-28 (10 months ago)
Commits, last 90 days0
Releases, last 12 months0
Contributors1
Issues closed, last 90 days0
Pull requests merged, last 90 days0
Project age11 months

Checked on 2026-10-17 with the GitHub API.

Overview

mcp-framework is a Rust framework for building AI agents that use MCP servers. It is built on the official Rust rmcp MCP implementation and Tokio, with support for connecting to multiple servers over HTTP and stdio, creating custom MCP tools, and inspecting them in a web UI.

The project’s agent integrations cover Anthropic Claude and OpenAI. It is aimed at Rust developers who want to combine LLMs with MCP tools in an async application, rather than assemble the client, server and agent pieces separately.

Key Features

  • Build MCP servers and register custom tools.
  • Connect an MCP client to multiple servers using HTTP and stdio transports.
  • Run async agents with conversation and session management.
  • Integrate Anthropic Claude and OpenAI models, including streaming support for Claude.
  • Inspect and test tools through a web UI that shows request and response details.
  • Use the Playwright MCP integration for browser automation.

Use Cases

  • Rust teams can expose internal APIs or business logic as MCP tools for AI agents.
  • Developers can prototype an agent that calls tools from several MCP servers.
  • Teams can inspect tool behavior and responses through the web inspector while developing an MCP server.
  • Developers can build browser-task agents using the Playwright MCP integration and a supported LLM.

What you need

Detected in the repository

  • Rust (from Cargo.toml)

License in plain words

MITpermissive

  • Commercial use: yes
  • Modify and redistribute: yes
  • You must keep: the copyright and license notice
  • Share your changes: no

A summary, not legal advice: the LICENSE file is what applies.

Getting Started

git clone https://github.com/koki7o/mcp-framework
cd mcp-framework
cargo run

Requires Rust 1.70 or later. See the README for agent examples, API key setup and inspector instructions.

Alternatives

  • java-sdk: The Java SDK provides MCP client and server APIs for Java applications, while mcp-framework targets Rust and adds agent integrations and a web inspector.
ProjectLanguageLicenseStarsStatus
mcp-frameworkRustMIT25Quiet
java-sdkJavaMIT3.7kActive

Considerations

The repository is marked Quiet: it has had no commits in the last 90 days and no release in over a year. Its measured contributor history is concentrated in one person, so teams should account for limited maintainer redundancy and verify compatibility with their chosen MCP servers and model APIs. The README lists vector databases, graph-based orchestration, MCP resources and prompts, authentication, and a memory system as planned rather than available features. Rust is required; the repository does not declare CI or tests in the supplied project facts. The MIT license permits broad reuse, subject to its license terms.

Found this useful?

Share it with someone who would like mcp-framework.