February 15, 2026
Klaw & Order: Agent Intent
Show HN: Klaw.sh – Kubernetes for AI agents
Big promises, bigger questions: is it open and what does it actually do
TLDR: Klaw pitches itself as “Kubernetes for AI agents,” promising easy, scaled orchestration of smart bots from one install. Commenters are split between excitement and confusion, debating whether it’s truly open-source, if it actually uses Kubernetes, and whether the bold tagline matches the real capabilities
“Kubernetes for AI agents” just dropped, and the comment section is a rollercoaster. Klaw claims a one‑file install that lets you run swarms of AI “agents” (think smart bots) from a laptop to a cluster, even piping them into Slack. Fans are jazzed at the idea of giving bots more power and scale. One early cheerleader even joked about the next step: “give the agent control of a whole data center.”
But then the mood split. Confused readers asked if it actually runs on Kubernetes (the popular system for managing apps across many computers) or just borrows the vibe. Some thought it was a Kubernetes add‑on; others realized it’s its own orchestrator and felt the tagline oversold it. The “what does this really do?” chorus grew louder as folks asked if it’s about pooling hardware, sandboxing agents, or something else entirely.
And yes, there’s drama: a commenter dropped a link to “Gas Town” and claimed, pointedly, that unlike those tools, this isn’t open source—clashing with the project’s own “open-source” framing. Meanwhile, the feature list sounds loaded: 300+ language models (via a single API), scheduled jobs, container isolation, and Slack bots. The crowd’s verdict? Ambitious idea, unclear messaging. Hype lovers see the future; skeptics smell marketing spin; everyone else wants a straight answer before handing the keys to their servers
Key Points
- •Klaw is an open-source platform to deploy, orchestrate, and scale AI agents, described as analogous to Kubernetes for agents.
- •It runs as a single binary with no dependencies and supports interactive CLI chat or a full platform with Slack bot and scheduler.
- •Configuration supports three provider options: each::labs Router (300+ models), OpenRouter, and direct Anthropic access.
- •Deployment modes include Single-Node, Distributed (controller-node), and Container Mode using Podman for isolation.
- •Features include Kubernetes-style clusters/namespaces, built-in tools, a skills system, cron-based scheduling, and multi-channel support (CLI, TUI, Slack).