An MCP server that lets AI agents run any Apify Store scraper or automation Actor to extract data from social media, search engines, maps, and websites.
stacklok/toolhive
An open source platform for running MCP servers in isolated containers, with access policies, a gateway, a registry, and a Kubernetes operator.
- Stars
- 2,326
- Forks
- 326
- Open issues
- 421
- Last commit
- 5 hours ago
What is it?
ToolHive is a platform for running and managing Model Context Protocol (MCP) servers. It runs each MCP server in an isolated container with a minimal permission profile, so servers do not need local credentials on the developer's machine. When configured with an authentication source, it enforces identity and access policy on each request.
The project is organized into three components: a Gateway that exposes team endpoints and centralizes authentication, auditing and tool filtering; a Registry Server that curates a catalog of approved servers; and a Runtime that runs servers locally via Docker or Podman or in Kubernetes. Users can work through a desktop UI or a CLI, and the Kubernetes Operator manages servers with custom resources, OIDC/OAuth SSO, OpenTelemetry and Prometheus metrics.
ToolHive suits developers who want safer local MCP setups, and platform teams that need to self-host MCP servers on existing Kubernetes infrastructure with audit logs and access control.
Key features
- 01Runs each MCP server in an isolated container with a minimal permission file
- 02Gateway with SSO via OIDC/OAuth, access policies and audit logging
- 03Registry Server for curating and signing approved MCP servers
- 04Runtime supports local Docker or Podman and Kubernetes deployment
- 05Desktop UI and CLI for discovering, configuring and running servers
- 06Kubernetes Operator with custom resources and OpenTelemetry and Prometheus support
Use cases
- Running MCP servers in Claude Code, Cursor or VS Code without storing credentials locally
- Self-hosting a governed MCP registry and gateway inside an enterprise network
- Deploying and auditing MCP servers across a Kubernetes cluster
Quick start
- 1
Download ToolHive from the official downloads page
- 2
Follow the CLI quickstart guide in the documentation
Things to know
- Local MCP servers run through Docker or Podman, so one of those must be installed.
- Some advanced capabilities are offered in Stacklok Enterprise rather than the open source project.
In collections
Topics
FAQ
- What is toolhive?
- An open source platform for running MCP servers in isolated containers, with access policies, a gateway, a registry, and a Kubernetes operator. ToolHive is a platform for running and managing Model Context Protocol (MCP) servers. It runs each MCP server in an isolated container with a minimal permission profile, so servers do not need local credentials on the developer's machine. When configured with an authentication source, it enforces identity and access policy on each request.
- Is toolhive free to use? What license does it have?
- toolhive is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
- Is toolhive still maintained?
- Actively maintained. The last commit to toolhive was on Nov 6, 2026.
- What can I use toolhive for?
- Running MCP servers in Claude Code, Cursor or VS Code without storing credentials locally Self-hosting a governed MCP registry and gateway inside an enterprise network Deploying and auditing MCP servers across a Kubernetes cluster
- How do I install toolhive?
- Download ToolHive from the official downloads page Follow the CLI quickstart guide in the documentation
- Where can I get toolhive?
- The source code and installation instructions are on GitHub: https://github.com/stacklok/toolhive
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
An email service for autonomous AI agents that lets them register, read, send, and manage their own @atomicmail.ai inbox via JMAP.
Open-source Privileged Access Management (bastion host) platform for secure access to SSH, RDP, Kubernetes, databases, and web apps.
Clair is an open source static analysis tool that indexes container images and matches them against known vulnerabilities.
A Go MCP server giving LLM clients web search, image search and page-to-Markdown scraping, with no API keys needed.
Open-source identity and access layer that connects AI agents to 1200+ external systems with unified auth, permissions and audit logs.