AppRisingofficial publisher

ToolHive

App : A product you use directly

Runs each MCP server in an isolated container with managed permissions and secrets.

The cleanest way to run third-party MCP servers without giving them access to your machine.

Official site GitHub · 2,306 ★

Overall score

8.7 /10

Security: nothing to flag

Utility

7.5

Reliability

10.0

Security

No caveat on record

Ease

5 min · Technical

Security reflects where data goes, requested permissions and incidents on record at the verification date. It is not a guarantee of zero risk: check the deployment mode and the connections you enable.

What it does, concretely

Running third-party MCP servers on your machine exposes your entire file system and credentials to tools whose code you do not control. ToolHive solves this by isolating each MCP server in its own container, with network permissions and secrets managed separately.

In practice, ToolHive acts as a gateway between your AI clients (Claude Code, Cursor, GitHub Copilot, VS Code) and MCP servers. It launches each server in a Docker or Podman container, enforces per-request access policies, encrypts secrets, and exposes a single unified endpoint. A built-in semantic search engine called MCP Optimizer dynamically selects only the relevant tools for each request, reducing token usage by 60 to 85 %.

ToolHive is aimed at individual developers who want to use MCP servers without putting their local environment at risk, at platform teams who want to deploy MCP on Kubernetes with audit logging and SSO, and at enterprises that cannot send their data through a third-party SaaS provider.

On the other hand, if you do not use any MCP servers or work in an environment without Docker or Podman, ToolHive offers little value. Advanced enterprise features such as dedicated support and a hardened distribution are part of the commercial Stacklok platform, which is separate from the open source tool.

What it offers

·Each MCP server runs in an isolated Docker or Podman container with no access to local credentials by default.
·A built-in registry lists official and community MCP servers vetted by the Stacklok team.
·The MCP Optimizer reduces token usage by 60 to 85 % by sending only the relevant tools to the model for each request.
·Secrets are encrypted and injected into containers without being exposed to the host system.
·A desktop graphical interface (ToolHive UI) lets you discover, configure, and launch MCP servers without using the terminal.
·A Kubernetes operator enables production deployment and management of MCP servers with access policies, OIDC/OAuth SSO, OpenTelemetry traces, and Prometheus metrics.
·The central gateway groups multiple MCP servers behind a single endpoint and integrates with compatible AI clients.
·Provenance verification and registry server signing allow you to control the origin of deployed images.

Our tips

Enable the MCP Optimizer from the start: activating semantic tool search significantly reduces the number of tokens consumed per request without changing your workflow.
Manage your secrets through the ToolHive interface rather than passing them as plain environment variables: they are encrypted and never exposed in clear text on the host system.
If you are just getting started, use the ToolHive UI desktop app to explore the registry and launch your first servers; switch to the CLI for automation and scripted deployments.
To test an MCP server before connecting it to your AI client, use the built-in MCP Inspector to validate its behaviour in an isolated environment.
On Kubernetes, declare your MCP servers using the ToolHive operator's Custom Resource Definitions to benefit from fleet management, access policies, and identity provider integration.

Watch out

ToolHive requires Docker or Podman to run locally: without one of these container engines installed and running, no MCP server can be launched.
Advanced enterprise features such as dedicated support, a hardened distribution, and additional capabilities are not part of the free open source tool and belong to the commercial Stacklok platform.

In practice

Price: Free
Commercial use: Commercial use: yes
Available on: cli, docker, macos, windows
Licence: Apache-2.0

Security and data

Rating: Security: nothing to flag
Your data: Data stays local
Maintenance: Active
Last release: 22 October 2026

Links and repository checked on 25 October 2026 · added on 1 May 2025

How to install ToolHive

1. Make sure Docker or Podman is installed and running on your machine before anything else. 2. On macOS or Linux, install the CLI with Homebrew: brew tap stacklok/tap then brew install thv. 3. On Windows, install the CLI with winget: winget install stacklok.thv. 4. For the graphical interface, download the ToolHive UI desktop app from stacklok.com/download for your system (macOS Apple Silicon, macOS Intel, Windows x64, Linux .deb or .rpm). 5. Launch ToolHive and browse the built-in registry to choose a vetted MCP server rather than importing one from an unknown source. 6. Configure your secrets through the ToolHive interface so they are encrypted and never transmitted in plain text on your system.

brew tap stacklok/tap && brew install thv

In the same category

Whole category Monitor →