Toolhive, developed by Stacklok, is a free MCP server management and orchestration platform for the mcp category. Built with Go, it is distributed as a single binary CLI, a desktop application for Windows, macOS, and Linux, and a Kubernetes Operator. This tool connects to and orchestrates Model Context Protocol (MCP) servers, which are essential for LLM-based applications and AI agents to access external data and services. It provides a centralized approach to deploying and managing these critical components across various environments.

What does Toolhive connect to?

Toolhive connects to and orchestrates Model Context Protocol (MCP) servers, which are used by large language model (LLM)-based applications and AI agents. These MCP servers enable AI to access external data and services. Toolhive functions as an MCP server management and orchestration platform. It is designed for deployment and management using Kubernetes and containers. The program acts as a control layer for MCP servers rather than directly exposing specific callable tools itself.

Core features and tools

Through its Virtual MCP Server (vMCP) component, Toolhive exposes composite tools, which are multi-step workflows executed across multiple backend MCP servers. It also provides a Registry Server that implements the official MCP Registry API, allowing discovery and access of managed MCP servers and agent skills. Toolhive supports stdio, Server-Sent Events (SSE), and Streamable HTTP transport protocols. Streamable HTTP offers richer interaction patterns and bidirectional streaming. Compatible AI clients include GitHub Copilot, Cursor, Claude Code, ChatGPT, and VS Code.

Who is it for?

Toolhive serves individual developers, development teams, and enterprises. Individual developers can manage local MCP servers using its desktop UI or CLI. Teams and enterprises can deploy and centrally manage MCP servers at scale within Kubernetes environments. Administrators can curate and govern a trusted catalog of MCP servers and skills via the Registry Server. It supports orchestrating multi-step workflows for complex operations. Skip this if you do not need to manage or orchestrate Model Context Protocol servers for AI applications.

How to get started

Access is granted through a standard OAuth authorization flow, connecting to existing OpenID Connect (OIDC) identity providers such as Google, GitHub, Okta, or Microsoft Entra ID. Toolhive separates authentication from authorization, requiring clients to present valid JWT tokens. Authorization is built on Amazon's Cedar policy language, needing an authorization configuration file.

Use cases

• Managing MCP servers: Individual developers can deploy local MCP servers via desktop UI or CLI, while teams can manage them at scale in Kubernetes using the Operator.
• Curating a trusted catalog: The Toolhive Registry Server enables administrators to build and govern a catalog of MCP servers and skills from various sources.
• Orchestrating multi-step workflows: Users can define and execute composite tools that orchestrate multi-step workflows across multiple backend MCP servers.

Toolhive provides a capable management and orchestration solution for Model Context Protocol servers, catering to various user scales from individual developers to enterprises. Its support for multiple deployment methods and transport protocols, combined with strong authentication and authorization, makes it a versatile tool for AI agent integration. The ability to define composite tools for multi-step workflows adds significant value. Start by exploring its capabilities for deploying and managing local MCP servers using the desktop UI or CLI.