You are running inside klaw, an open-source AI agent orchestration platform. This document explains your capabilities and how to use them effectively.
$npx -y skills add klawsh/klaw.sh --skill klaw.shInstalls into the current project.
Run `npx skills use "https://github.com/klawsh/klaw.sh" --skill "klawsh/klaw.sh"` and follow the generated skill instructions now. Read its complete output, redirecting it to a temporary file first if necessary. Resolve relative paths from the supporting-files directory it provides.
Use the skills in "https://github.com/klawsh/klaw.sh" that are relevant to the current task. Run `npx skills add "https://github.com/klawsh/klaw.sh"` and select the relevant skills, then follow their instructions.
| 1 | <p align="center"> |
| 2 | <img src="docs/assets/klaw-logo.png" alt="klaw" width="140" /> |
| 3 | </p> |
| 4 | |
| 5 | <h1 align="center">klaw</h1> |
| 6 | |
| 7 | <p align="center"> |
| 8 | <strong>kubectl for AI Agents</strong> |
| 9 | </p> |
| 10 | |
| 11 | <p align="center"> |
| 12 | Enterprise AI agent orchestration. Manage, monitor, and scale your AI workforce.<br/> |
| 13 | One binary. Deploys in seconds. Scales to hundreds of agents. |
| 14 | </p> |
| 15 | |
| 16 | <p align="center"> |
| 17 | <a href="https://klaw.sh/docs">Documentation</a> • |
| 18 | <a href="#what-is-klaw">What is klaw?</a> • |
| 19 | <a href="#quick-start">Quick Start</a> • |
| 20 | <a href="#slack-control">Slack Control</a> • |
| 21 | <a href="#architecture">Architecture</a> |
| 22 | </p> |
| 23 | |
| 24 | <p align="center"> |
| 25 | <img src="https://img.shields.io/badge/license-each::labs-green.svg" alt="License" /> |
| 26 | <img src="https://img.shields.io/badge/go-1.24+-00ADD8.svg" alt="Go Version" /> |
| 27 | <img src="https://img.shields.io/github/stars/klawsh/klaw.sh?style=social" alt="Stars" /> |
| 28 | <a href="https://deepwiki.com/klawsh/klaw.sh"><img src="https://deepwiki.com/badge.svg" alt="Ask DeepWiki"></a> |
| 29 | </p> |
| 30 | |
| 31 | --- |
| 32 | |
| 33 | ## What is klaw? |
| 34 | |
| 35 | **klaw** is enterprise AI agent orchestration — like kubectl, but for AI agents. |
| 36 | |
| 37 | ```bash |
| 38 | # See all your agents |
| 39 | $ klaw get agents |
| 40 | |
| 41 | NAME NAMESPACE STATUS MODEL LAST RUN |
| 42 | lead-scorer sales running claude-sonnet-4 2m ago |
| 43 | competitor-watch research idle gpt-4o 1h ago |
| 44 | ticket-handler support running claude-sonnet-4 30s ago |
| 45 | report-gen analytics idle claude-sonnet-4 6h ago |
| 46 | |
| 47 | # Check what an agent is doing |
| 48 | $ klaw describe agent lead-scorer |
| 49 | |
| 50 | Name: lead-scorer |
| 51 | Namespace: sales |
| 52 | Status: Running |
| 53 | Model: claude-sonnet-4-20250514 |
| 54 | Skills: crm, web-search |
| 55 | Tools: hubspot, clearbit, web_fetch |
| 56 | Last Run: 2 minutes ago |
| 57 | Next Run: in 58 minutes (cron: 0 * * * *) |
| 58 | |
| 59 | # View real-time logs |
| 60 | $ klaw logs lead-scorer --follow |
| 61 | |
| 62 | [14:32:01] Fetching new leads from HubSpot... |
| 63 | [14:32:03] Found 12 new leads |
| 64 | [14:32:05] Analyzing lead: john@acme.com |
| 65 | [14:32:08] Score: 85/100 (Enterprise, good fit) |
| 66 | [14:32:09] Updated HubSpot lead score |
| 67 | ... |
| 68 | ``` |
| 69 | |
| 70 | ### Or control everything from Slack |
| 71 | |
| 72 | ``` |
| 73 | You: @klaw status |
| 74 | |
| 75 | klaw: 📊 Agent Status |
| 76 | ├── lead-scorer (sales) — running, 2m ago |
| 77 | ├── competitor-watch (research) — idle |
| 78 | ├── ticket-handler (support) — running, 30s ago |
| 79 | └── report-gen (analytics) — idle |
| 80 | |
| 81 | You: @klaw run competitor-watch |
| 82 | |
| 83 | klaw: 🚀 Starting competitor-watch... |
| 84 | Checking competitor.com/pricing... |
| 85 | Found 2 pricing changes since yesterday. |
| 86 | Posted summary to #competitive-intel |
| 87 | ``` |
| 88 | |
| 89 | --- |
| 90 | |
| 91 | ## The Problem |
| 92 | |
| 93 | You're running AI agents in production: |
| 94 | - **Lead Scorer** — analyzes CRM leads every hour |
| 95 | - **Competitor Watch** — monitors competitor websites daily |
| 96 | - **Ticket Handler** — auto-responds to support tickets |
| 97 | - **Report Generator** — creates weekly analytics reports |
| 98 | |
| 99 | But managing them is chaos: |
| 100 | |
| 101 | | Challenge | Current State | With klaw | |
| 102 | |-----------|---------------|-----------| |
| 103 | | **Visibility** | "Is the agent running? What's it doing?" | `klaw get agents`, `klaw logs` | |
| 104 | | **Isolation** | Sales agent accessing support secrets | Namespaces with scoped permissions | |
| 105 | | **Scheduling** | Messy cron jobs, Lambda functions | `klaw cron create` — built-in | |
| 106 | | **Scaling** | Manual server provisioning | `klaw node join` — auto-dispatch | |
| 107 | | **Debugging** | grep through CloudWatch | `klaw logs agent --follow` | |
| 108 | | **Deployment** | Complex setup, many dependencies | Single binary, one command | |
| 109 | |
| 110 | **OpenClaw works, but deployment is painful and scaling is worse.** |
| 111 | |
| 112 | klaw brings Kubernetes-style operations to AI agents. One binary. Deploys in seconds. |
| 113 | |
| 114 | --- |
| 115 | |
| 116 | ## Quick Start |
| 117 | |
| 118 | ### 1. Install |
| 119 | |
| 120 | ```bash |
| 121 | curl -fsSL https://klaw.sh/install.sh | sh |
| 122 | ``` |
| 123 | |
| 124 | ### 2. Configure |
| 125 | |
| 126 | ```bash |
| 127 | # Pick one provider |
| 128 | export ANTHROPIC_API_KEY=sk-ant-... # Direct Anthropic |
| 129 | export OPENROUTER_API_KEY=sk-or-... # OpenRouter (100+ models) |
| 130 | export EACHLABS_API_KEY=... # each::labs (300+ models) |
| 131 | ``` |
| 132 | |
| 133 | ### 3. Run |