Skip to main content

​
Prerequisites

  • A terminal (macOS, Linux, or Windows WSL)
  • An API key from one of the supported providers

​
Step 1: Install klaw

Verify the installation: 
klaw version

​
Step 2: Configure Your Provider

klaw supports multiple LLM providers. Choose one and set your API key:

​
Step 3: Start Chatting

klaw chat
You’ll see an interactive prompt where you can start conversing with the AI agent: 
klaw v1.0.0 - The Kubernetes for AI Agents
Using model: claude-sonnet-4-20250514

> Hello! Can you help me understand this codebase?

I'll help you explore this codebase. Let me start by looking at the
project structure...

[Tool: glob] Finding files...
├── cmd/
├── internal/
├── go.mod
└── README.md

This appears to be a Go project. Let me read the main files to give
you a better overview...

​
Step 4: Try Some Commands

​
Interactive Chat with TUI

For a richer experience with syntax highlighting and better formatting: 
klaw chat --tui

​
One-shot Task

Execute a single task and exit: 
klaw dispatch "Explain what this project does"

​
Specify a Model

Use a specific model: 
klaw chat --model claude-sonnet-4-20250514
klaw chat --model gpt-4o

​
Start Full Platform

Run the full platform with Slack integration and scheduler: 
export SLACK_BOT_TOKEN=xoxb-...
export SLACK_APP_TOKEN=xapp-...
klaw start

​
What’s Next?

Understand Agents

Learn how agents work and how to configure them

Explore Tools

Discover the built-in tools agents can use

Slack Integration

Deploy your agent to Slack

Distributed Mode

Scale across multiple machines

​
Example Use Cases

> Fix the bug in src/api/handler.go where users can't authenticate

# Agent reads the file, identifies the issue, and fixes it

> Research the latest developments in WebAssembly and summarize them

# Agent searches the web, reads multiple sources, and provides a summary

> Generate API documentation for all the handlers in internal/api/

# Agent reads the code and generates comprehensive documentation

> Check the status of all Docker containers and restart any that are unhealthy

# Agent runs docker commands, analyzes output, and takes action