Stacklok
Enterprise MCP operations on Kubernetes with registry, runtime, gateway, policy, and observability controls.
About Stacklok
Stacklok is an enterprise platform for running AI agents and Model Context Protocol servers on existing Kubernetes infrastructure. It treats MCP servers as pods and uses familiar cluster boundaries, networking rules, service meshes, and GitOps workflows to control how agents reach business systems and data. Teams can curate trusted servers in a central registry, deploy and manage them through a governed runtime, expose tools through one gateway, and give administrators and knowledge workers access through a portal. Native OpenTelemetry instrumentation sends tool-call activity to an existing observability stack. Identity controls map Kubernetes ServiceAccounts and OIDC claims to MCP permissions. Stacklok is built on ToolHive, so teams can evaluate the open-source foundation before adopting the production platform. This approach fits organizations that already operate Kubernetes and want MCP governance to sit alongside their current platform controls, but it also brings cluster operations, policy design, and enterprise procurement into the adoption path.
Best for
Best use cases
Replace unmanaged MCP usage with a centrally governed registry and runtime
Apply existing Kubernetes and GitOps controls to MCP servers
Send agent tool-call telemetry into an established observability stack
Compare before choosing
Compared with local-only MCP setups, Stacklok adds centralized deployment and oversight
Compared with a separate agent runtime, it reuses Kubernetes infrastructure and controls
Compared with open-source ToolHive evaluation, the enterprise platform is positioned for production operation
Features
Govern MCP servers on Kubernetes
Deploy and manage MCP servers as Kubernetes workloads while applying namespaces, ingress, network policy, and service mesh controls.
Curate trusted agent tools
Maintain a central registry where teams can discover approved hosted and local MCP servers.
Observe and control agent access
Route tools through a gateway, map workload identities to permissions, and export tool-call telemetry through OpenTelemetry.
Pros
- Reuses established Kubernetes controls, which can reduce platform duplication.
- Central discovery and access workflows help curb unmanaged MCP server use.
- Tool-call telemetry can flow into an existing OpenTelemetry-compatible stack.
Cons
- Teams need Kubernetes operations skills to fit the platform into production.
- Contact-based subscription pricing slows direct cost comparison.
- Production rollout requires policy, identity, and GitOps integration work.
Pricing Plans
Enterprise
- Curated MCP server registry
- Kubernetes-based MCP runtime
- Unified tool gateway
- Administrative and user portal
- Policy-as-code and identity integration
- OpenTelemetry-based oversight
* Pricing information is for reference only. Please refer to the latest announcement on Stacklok's official website.