Clawdbot has been renamed Moltbot. The project is built around a simple idea: run a personal AI assistant on your own computer, then connect it to messaging apps such as WhatsApp, Telegram, Discord, Slack, and iMessage so conversations with AI can happen inside the chat tools people already use every day.
According to the source material, Moltbot supports local data storage, voice interaction, and multi-platform messaging access, while also working with models including Claude, ChatGPT, and Gemini. The project has collected more than 11,800 GitHub stars and over 1,500 forks. The software itself is open source and free to use, while model usage costs depend on the AI provider a user chooses.
One assistant across multiple messaging platforms
The main appeal is not a standalone interface but a shared control layer. After setup, the same assistant can appear across several communication platforms at once. The article lists WhatsApp, Telegram, Discord, Slack, Microsoft Teams, Signal, Google Chat, and macOS-only iMessage among the supported services.
That gives Moltbot a different role from a standard chatbot app. Instead of opening a separate AI window, users can ask questions, organize tasks, or send commands directly inside their existing chats. It is a practical model for people who already move between several messaging platforms during the day. Fewer app switches. That is the pitch.
Local-first design comes with setup overhead
Moltbot positions local deployment as a privacy feature. The source says conversation data stays under the user’s control rather than being stored by a third-party service by default, which is a key distinction from fully cloud-based AI products.
There is also a clear tradeoff. Installation requires some command-line familiarity. The published system requirements specify Node.js 22 or later, with support for macOS and Linux, while Windows users need WSL2. The recommended installation path uses npm or pnpm, followed by clawdbot onboard --install-daemon to run the onboarding wizard and install the Gateway daemon for persistent operation.
The guide also includes basic commands for launching the Gateway, sending a message, and invoking the agent to perform tasks. For developers or power users, this will look manageable; for casual users, it is a real barrier to entry.
Voice features, skills, and automation layers
Moltbot is not limited to text prompts. The source describes voice wake, ongoing spoken conversation, and device support tied to Mac, iPhone, and Android nodes. Optional companion apps include a macOS app, iOS Node, and Android Node, with functions such as menu bar Gateway control, WebChat, remote management, camera access, screen recording, and screen capture depending on the device.
The project also includes a skills framework. Its workspace root is listed as ~/clawd, with behavior shaped through prompt files such as AGENTS.md, SOUL.md, and TOOLS.md. Skills can be extended under ~/clawd/skills/<skill-name>/SKILL.md. ClawdHub is described as a skill registry that lets the AI discover and import new capabilities.
Automation options include Cron scheduling, Webhooks, and Gmail Pub/Sub triggers. Browser control is also part of the stack, using a dedicated Chrome or Chromium instance for snapshots, actions, uploads, and profile management. At that point, Moltbot starts to look less like a chat add-on and more like a personal automation hub.
Each platform has its own connection path
Cross-platform support does not mean a single setup flow. WhatsApp requires clawdbot channels login to link a device. Telegram depends on a TELEGRAM_BOT_TOKEN. Slack needs both a Bot Token and an App Token. Discord uses a Bot Token, Signal requires signal-cli, and Microsoft Teams needs a Teams App plus Bot Framework configuration.
So the polished part comes later. The front-loaded work is still fairly technical. The comparison table in the source makes this point directly: unlike ChatGPT App or Claude App, Moltbot offers messaging-app integration, multi-platform support, local data handling, and open-source availability, but it does demand some technical skill during installation.
Security controls focus on pairing, allowlists, and sandboxing
The source gives notable space to security controls. In the default DM pairing mode, unknown senders receive a pairing code and the bot does not process their messages automatically. Users can also restrict access through allowFrom allowlists. Non-primary conversations in groups or channels can run inside a Docker sandbox.
A diagnostic tool, clawdbot doctor, is included to help surface potential security risks. At the same time, the article warns that this is still a relatively new project with broad permission requirements, so deployment should be handled carefully. It also points readers to a separate warning that poor configuration could expose serious vulnerabilities, including cases tied to crypto wallet losses.
Viewed as a whole, Moltbot is not a lightweight chatbot wrapper. It is a local-first, extensible personal AI assistant framework aimed at users who want AI embedded inside their communication stack, while keeping control of data and configuration on their own systems.