CrewClaw

CrewClaw

CrewClaw · Coding · Productivity

CrewClaw is an AI agent development platform with 199 ready-made AI agent templates you can deploy in about five minutes. You drop the files into your own machine and run them with a single command. Don't want to touch a server? Pay the team to host your AI employees for you. Each agent is defined in plain markdown files, so you can read, edit, and repurpose them without learning a framework.

Interface preview of CrewClaw

About CrewClaw

What Is CrewClaw

CrewClaw is a self-hosted AI agent platform built around a simple idea: an agent is just a set of markdown files. Instead of locking you into a visual builder or a proprietary cloud, it ships the actual configuration files, including persona, memory, skills, tools, and channels, so you own the whole thing forever.

It targets solo developers and small teams who want AI agents doing real work without paying per-seat SaaS fees. The library covers engineering, content, data, sales, operations, design, legal, and DevOps, with pre-configured teams that coordinate several agents on one job.

The main limit is that you bring your own model key. CrewClaw runs on Anthropic, OpenAI, or a local Ollama model, and your usage costs are separate from what you pay for the platform.

Getting Started

  1. Buy a self-host tier ($9 one-time) or a hosted tier ($29/mo) from the CrewClaw site.
  2. Pick your agents from the 199-item library, or swap any of them within the first seven days.
  3. Download the deploy package, drop the agent folder into your project, and run npx openclaw start.
  4. Add your Anthropic or OpenAI API key so the agent has a model to think with.
  5. Connect a channel like Telegram, Slack, or Discord and send your first message.

Product Information

A quick look at CrewClaw's pricing, supported platforms, and performance.

Free PlanNo
Paid Plans$9 one-time - $29/mo
PlatformWeb, Desktop (self-hosted), Telegram, Slack, Discord
DeveloperCrewClaw
CategoryCoding · Productivity
Release DateJan 2025
Latest UpdatedSep 2025
Website Visits8K
Website Global Rank2.7M
API AvailabilityYes

Best for

The users, tasks, and scenarios where this tool fits best.

Users

  • Solo founders
  • Small dev teams
  • Content marketers

Tasks

  • SEO content pipelines
  • Automated code review
  • Customer support

Scenarios

  • Replacing a repeated weekly chore
  • Tracking competitors
  • Running a freelance business

Key features

199 Production-Ready Agent Configs

The library spans eight workflow areas, from 38 engineering agents to 14 design agents. Every config is a working setup, not a blank template, so you can deploy it as-is or open the markdown and adjust the persona, skills, and tools. It's the fastest part of the whole product. No prompt tuning. No blank page.

Multi-Agent Teams That Coordinate

CrewClaw isn't only single agents. Teams of three or four agents share context and split a job, like a content crew where one agent does keywords, another drafts, and a third schedules posts. The coordination config lives in an AGENTS.md file you can read and rewrite. Easy to extend.

Markdown File Architecture

Each agent is defined by files you can open in any text editor: a SOUL.md for persona, a MEMORY.md for persistent state, and modular skill and tool files. That design means no vendor lock-in. If you stop paying, the files still work. The same architecture Nvidia calls the core of every company's stack ships pre-configured here.

Model Flexibility

Agents run on Claude, GPT, or a local Ollama model. You bring your own API key, so you control both the cost and which model handles which task. Want a cheap model for drafts and a stronger one for final reviews? That's a config change, not a platform decision. Swap it in one line.

Any Channel, Including Telegram

CrewClaw's agents go live where you already talk: Telegram, Slack, and Discord. The hosted tier provisions a Telegram bot for you, which is why the pitch is "live on your Telegram in minutes." These AI agents for Telegram work like a coworker you message. You type a request, and the agent keeps working in the background.

Self-Host or Hosted

The self-host tier is a one-time $9 and yours forever. The hosted tier is $29/mo, and the team runs the agent around the clock for you with no server to maintain. If you'd rather keep your data close, self-hosted AI agents run entirely on your machine. Both come with a 30-day refund and a 7-day swap window on your agent picks.

Pros and cons

Pros

  • One-time $9 self-host option means no recurring platform fee and no vendor lock-in.
  • Agents are plain markdown, so you can audit and edit every part of how they behave.
  • 199 configs across eight categories save the prompt-engineering grind of building from scratch.
  • Multi-agent teams handle whole workflows, not just single repeated tasks.
  • Works with Claude, GPT, or a local Ollama model, so you're not tied to one provider.

Cons

  • No free plan, so you can't test the configs before paying; the 30-day refund is your safety net.
  • You must supply your own model API key, which adds usage costs on top of the platform price.
  • The hosted tier is $29/mo, which adds up for anyone who could run it themselves for $9 once.

Frequently asked questions

It's a set of markdown files that define a persona, memory, skills, tools, and channels. Run it and it behaves as a focused AI employee: a code reviewer, an SEO specialist, a support agent. You keep it in your own project.