← Back to Discover

ToolHive

Open-source platform that runs Model Context Protocol servers in isolated containers, with a desktop app, a CLI and a Kubernetes operator, plus a gateway and registry for teams.

MCP serversAI security & guardrails
Visit website ↗

What it does

ToolHive is an open-source platform for running and managing MCP (Model Context Protocol) servers. Each server runs in its own container with a minimal permission file and no local credentials, and ToolHive can connect the servers to AI clients such as Claude Code, Cursor, GitHub Copilot and VS Code. Individual developers can use the desktop UI or the CLI to discover, configure and run servers locally through Docker or Podman. It also offers encrypted secret storage, network isolation controls, the official MCP Inspector for testing, and an MCP Optimizer aimed at reducing token usage. For teams, a Kubernetes operator manages servers and registries centrally, with SSO through OIDC/OAuth, audit logging, OpenTelemetry and Prometheus metrics. A gateway applies access policies, and a registry server keeps a catalogue of approved servers, so the whole setup can be self-hosted instead of passing data through a SaaS provider.

Capabilities

  • Runs each MCP server in an isolated container
  • Desktop app and CLI for discovering, configuring and running servers
  • Gateway with access policies, SSO and tool filtering
  • Registry server for a curated catalogue of trusted servers
  • Kubernetes operator with audit logging and observability
  • MCP Optimizer to cut token usage

When to use it

  • You want MCP servers to run in containers instead of directly on your machine
  • Your platform team needs central policy, SSO and audit logs for MCP on Kubernetes
  • Compliance rules mean your MCP registry and gateway must be self-hosted