Stacklok Introduces Cloud-Native Harnesses for Reliable, Scalable AI Agents

What changed?

Stacklok launched a set of cloud-native infrastructure components to move AI agent harnesses beyond desktop-centered deployments. Mecatl, an open source project, separates the agent control loop from the client, model provider, state store, and execution environment. This allows agents to be managed as distributed systems in the cloud. ToolHive, also open source, extends agent management with a Kubernetes-based platform for deploying and governing agentic MCP servers in enterprise environments. Stacklok also introduced a commercial product, the enterprise spine, for centralized management (identity, policy, audit) across these components. AI Gateway, another component, manages access and routing to LLMs, with open-sourcing planned but not yet available.

Why does it matter to an everyday developer?

Desktop-first agent harnesses, even if shifted to server VMs or containers, struggle when scaled: issues with lifecycle management, session reliability, isolation, and enterprise control are common. By decoupling agent loops and adopting cloud-native design, these new Stacklok tools enable developers and teams to deploy, manage, and govern agents in centralized, scalable environments—especially important for regulated, large-scale enterprise contexts. This shift mirrors how containerization and orchestration (like Kubernetes) transformed traditional app deployment from desktops to scalable cloud platforms.