Now in public alpha

Kubernetes
for AI Agents

Deploy, orchestrate, and scale AI agents across your infrastructure.One binary. No dependencies. From laptop to cluster.

curl -fsSL https://klaw.sh/install.sh | sh
klaw chat
klaw start
klaw dispatch "task"

Everything You Need

From local development to production clusters, klaw handles the infrastructure so you can focus on your agents.

Single Binary

One ~20MB binary. No Python, no Docker, no dependencies. Just download and run.

kubectl-style CLI

Familiar commands: get, create, delete, describe, dispatch.

300+ LLM Models

Claude, GPT-4, Gemini, Llama, DeepSeek via each::labs Router or OpenRouter.

Multi-Channel

Slack, CLI, TUI, REST API. Deploy agents where your team works.

Distributed Mode

Controller-node architecture. Scale from laptop to enterprise cluster.

Namespaces

Kubernetes-style multi-tenancy. Isolate teams, projects, and environments.

Cron Jobs

Schedule recurring agent tasks. Automate reports, monitoring, and workflows.

Skills System

Composable capability bundles. web-search, git, docker, database, and more.

Container Runtime

Run agents in isolated Podman containers for security and reproducibility.

Kubernetes-Style Namespaces

Inspired by Kubernetes, klaw brings familiar multi-tenancy patterns to AI agent orchestration. Isolate teams, projects, and environments with first-class namespace support.

Team Isolation

Each team gets their own namespace with isolated agents, secrets, and resources. No cross-contamination.

$ klaw create namespace team-backend
$ klaw create namespace team-frontend
$ klaw create namespace team-data

Environment Separation

Run dev, staging, and production environments on the same cluster with complete isolation.

$ klaw create namespace dev
$ klaw create namespace staging
$ klaw create namespace prod
$
# Deploy to specific environment
$ klaw dispatch -n prod "analyze logs"

Resource Quotas

Set limits per namespace to control costs and prevent runaway agents from consuming all resources.

# Coming soon
$ klaw set quota team-ml --max-tokens 1M/day
$ klaw set quota team-ml --max-agents 10

RBAC Integration

Fine-grained access control. Users can only access namespaces they are authorized for.

# Switch namespace context
$ klaw config set-context --namespace=team-backend
$
# List agents in current namespace
$ klaw get agents

klaw vs Kubernetes

klaw borrows the best ideas from Kubernetes—namespaces, declarative configs, kubectl-style CLI—but is purpose-built for AI agents. No container orchestration overhead.

FeatureklawKubernetes
Installation
Single binary (~20MB)
Complex multi-component setup
Learning Curve
Minutes to get started
Weeks/months to master
Resource Usage
~50MB RAM idle
2GB+ for control plane
Namespaces
Built-in multi-tenancy
Built-in multi-tenancy
CLI Experience
kubectl-inspired, familiar
kubectl (the original)
Declarative Config
YAML manifests
YAML manifests
Focus
AI Agents & LLMs
Container orchestration
Built-in LLM Support
300+ models ready
Requires custom setup

Why not just use Kubernetes? Kubernetes is amazing for container orchestration, but it is overkill for AI agents. klaw gives you the patterns you love—namespaces, declarative YAML, familiar CLI—without the complexity of running a full K8s cluster. Think of it as "Kubernetes for AI Agents."

Quick Start in 30 Seconds

One binary. No dependencies. No configuration needed.

terminal — zsh

Three Deployment Modes

From local development to enterprise clusters. Same commands, same agents.

Single-Node

Local Development

klaw chat
klaw start

Interactive chat or full platform with Slack integration. Zero config required.

Distributed

Scale Across Machines

klaw controller start
klaw node join :9090
klaw dispatch "task"

Controller-node architecture. Dispatch tasks across multiple machines.

Containers

Isolated Execution

klaw build
klaw run agent --task
klaw ps

Run agents in Podman containers for security and reproducibility.

Powered by
ClaudeGPT-4GeminiLlamaDeepSeek+300 more

via each::labs LLM Router

Ready to deploy?

Get started in under a minute. No account required.

curl -fsSL https://klaw.sh/install.sh | sh
Copy