Overview
Clawdbot is an open-source, self-hosted AI assistant that connects large language models like Claude and GPT to your existing messaging platforms. Built by Peter Steinberger and the community, it runs entirely on your own hardware—whether macOS, Linux, Windows (WSL2), or even a small cloud instance—prioritizing data privacy and local control over cloud-dependent alternatives.
The platform routes conversations from WhatsApp, Telegram, Discord, Slack, Signal, iMessage, and more through a single WebSocket gateway. This architecture lets you manage multiple AI agents across different channels, automate recurring tasks with cron jobs, and execute shell commands or browse the web (with appropriate tool permissions and security configurations)—all while keeping your data local. It's designed for developers, power users, and teams who want full ownership of their AI infrastructure without vendor lock-in.
Clawdbot has gained significant traction with approximately 19,000-20,000 GitHub stars (as of early 2026) and praise from tech reviewers like Federico Viticci of MacStories, who called it "the ultimate expression of a new generation of malleable software that is personalized and adaptive." If you're comfortable with Node.js and command-line tools, Clawdbot delivers a production-ready personal assistant that adapts to your workflow rather than forcing you into a hosted service.
Key Features
Multi-Platform Messaging Integration — Connect WhatsApp, Telegram, Discord, Slack, Google Chat, Signal, iMessage, Microsoft Teams, and more to a single AI gateway, routing conversations to specialized agents without switching apps or managing separate bots.
Local-First Gateway Architecture — Run the WebSocket control plane on your own hardware (Mac mini, Linux VPS, or local machine) while pairing remote devices for camera, screen capture, or voice commands, ensuring your data never leaves your infrastructure by default.
Voice Wake & Talk Mode — Enable always-on speech recognition on macOS, iOS, or Android with push-to-talk or wake-word activation, powered by ElevenLabs for natural voice synthesis and continuous conversation without typing.
Dynamic Skills System — Install bundled, managed, or custom workspace skills (Markdown-based tool definitions) on the fly, letting the agent search a registry (ClawdHub), pull new capabilities automatically, and extend functionality without code changes.
Browser & Canvas Control — Launch and control dedicated Chrome/Chromium instances for web automation, render live agent-driven UI canvases with A2UI architecture, and execute visual tasks like screenshot capture or interactive debugging.
Cron Jobs & Automation — Schedule recurring tasks (reminders, reports, daily briefings) with cron syntax, trigger actions via webhooks and external integrations, and coordinate multi-agent workflows with session-to-session messaging (
sessions_sendtool).
Pricing & Plans
Clawdbot itself is completely free and open-source under the MIT license. You can self-host it on any machine with Node.js 22+ at zero cost, paying only for the infrastructure you choose (your own hardware or a $5/month VPS) and the LLM API usage (charged by your selected provider such as Anthropic or OpenAI based on API consumption, not subscription fees).
If you prefer a hosted solution without managing servers, ClawCloud Run offers managed hosting:
| Plan | Price (Monthly) | Price (Annual) | CPU/RAM | Disk/Traffic | Credits |
|---|---|---|---|---|---|
| Free | $0/month | — | Up to 4 vCPU/8GB | 10GB disk | $5 free |
| Hobby | $5/month | $60/year | Up to 16 vCPU/32GB | Unlimited | — |
| Pro | $20/month | $240/year | Up to 32 vCPU/64GB | Unlimited | — |
Notes:
- The core Clawdbot project has no licensing fees or subscription requirements.
- Self-hosting costs depend on your choice of hardware (local machine, cloud VPS, or Raspberry Pi-class devices).
- LLM costs are billed separately based on API usage by your chosen provider (e.g., Anthropic, OpenAI, or other compatible APIs according to their per-token pricing).
- ClawCloud Run pricing is for managed infrastructure only and is optional.
Best For
- Developers building personal automation workflows — If you want to script shell commands, automate cron jobs, or integrate AI into your existing tools without paying for cloud seats, Clawdbot delivers full programmatic control over your assistant.
- Privacy-conscious teams or individuals — Organizations handling sensitive data (legal, healthcare, finance) can run Clawdbot entirely on-premises or in a private VPC, never sending conversations to third-party hosted platforms.
- Power users managing multiple messaging platforms — Route WhatsApp family chats, Telegram work groups, and Discord communities to different agent workspaces with isolated contexts, all from a single gateway dashboard.
- Researchers or hobbyists experimenting with multi-agent systems — The
sessions_*tools let agents message each other, spawn subtasks, and coordinate work across sessions, ideal for exploring agent-to-agent orchestration patterns. For open-source alternatives, see Mastra AI. - Teams needing custom integrations — The skills system and WebSocket API allow you to extend Clawdbot with proprietary tools, internal databases, or company-specific workflows without forking the codebase.
FAQ
Is Clawdbot completely free to use?
Yes, the core Clawdbot project is open-source under the MIT license with no cost. You only pay for infrastructure (if you choose cloud hosting instead of a local machine) and LLM API usage charged by providers like Anthropic or OpenAI based on per-token consumption (not subscription fees). The optional ClawCloud Run managed hosting starts at $0/month with free credits.
Can I use Clawdbot without technical skills?
Clawdbot is designed for developers and technically proficient users. Setup requires installing Node.js 22+, running terminal commands, and configuring YAML files. The onboarding wizard (clawdbot onboard) streamlines setup, but you'll need basic command-line literacy. Non-technical users may prefer hosted alternatives like Retell AI or commercial AI chatbots.
Which messaging platforms does Clawdbot support?
Clawdbot integrates with WhatsApp (via Baileys), Telegram (grammY), Discord (discord.js), Slack (Bolt), Google Chat, Signal (signal-cli), iMessage (macOS only), BlueBubbles, Microsoft Teams, Matrix, Zalo, and WebChat. Each channel has configurable allowlists, DM pairing policies, and group routing rules documented at docs.clawd.bot/channels.
How does the DM pairing security model work?
By default, Clawdbot uses dmPolicy="pairing" for Telegram, WhatsApp, Signal, and other direct message channels. When an unknown sender messages the bot, they receive a pairing code and their message is not processed. You approve senders with clawdbot pairing approve <code>, adding them to a local allowlist. To allow open DM access, set dmPolicy="open" and include "*" in the channel allowlist (not recommended without additional authentication).
Can I run Clawdbot on a Raspberry Pi or low-power device?
Yes, Clawdbot runs on any system with Node.js 22+, including ARM-based devices like Raspberry Pi 4 or 5. Performance depends on your LLM provider (API calls happen server-side) rather than local compute. The Gateway itself is lightweight, but enabling browser automation and Canvas features may require more RAM (4GB minimum suggested based on community experience for basic usage; requirements vary by enabled tools).
What LLM models and providers are supported?
Clawdbot supports Anthropic (Claude Pro/Max with Opus 4.5), OpenAI (ChatGPT, GPT-5.2, Codex), Google Gemini, and any OpenAI-compatible API endpoint. The documentation recommends Anthropic Pro/Max for long-context tasks and better prompt-injection resistance. You configure models, failover policies, and API keys in the ~/.clawdbot/clawdbot.json config file. For voice capabilities, see our guide on AI voice generators.
How do I update Clawdbot to the latest version?
The recommended approach is to re-run the installation script or use clawdbot update --channel stable (or beta/dev for pre-release versions) to pull the latest release. After updating, run clawdbot doctor to check for configuration issues, migration tasks, or deprecated settings. If you installed from source, use git pull && pnpm install && pnpm build. Detailed upgrade steps are documented at docs.clawd.bot/install/updating.
Can I use Clawdbot for commercial or team use?
Yes, the MIT license allows commercial use without restrictions. For team deployments, consider running the Gateway on a shared Linux instance and using Docker sandboxing (agents.defaults.sandbox.mode: "non-main") for non-owner sessions to isolate untrusted group chats. You can also restrict tool access per agent (e.g., deny browser or cron for guest sessions) via the config file.
