Discover Moltbot (formerly Clawd AI), the viral self-hosted AI assistant with 60,000+ GitHub stars. Learn features, security, installation, and why it's revolutionizing agentic AI in 2026.
Dev Kraken
In early 2026, an open-source project called Clawd Bot (also known as Clawdbot) took the tech world by storm, amassing over 60,000 GitHub stars within just 72 hours of its viral launch. This unprecedented growth made it one of the fastest-growing open-source projects in GitHub history. On January 27, 2026, following a trademark request from Anthropic, the project was rebranded to Moltbot—but its revolutionary capabilities remain unchanged.
Unlike traditional AI chatbots that simply respond to queries, Moltbot represents a paradigm shift: it's an autonomous AI agent that actually does things. From booking flights and managing emails to running shell commands and automating browser tasks, Moltbot has been described as the closest thing we have to JARVIS from the Marvel universe.
In this comprehensive guide, we'll explore what makes Moltbot so special, how it works, its key features, security considerations, and whether it's right for you.
Moltbot is a self-hosted, open-source AI personal assistant designed to run on your own hardware—whether that's a Mac Mini, Windows PC (via WSL2), Linux machine, Raspberry Pi, or cloud server. Created by Austrian engineer Peter Steinberger, founder of PSPDFKit, Moltbot connects powerful AI language models like Anthropic's Claude, OpenAI's GPT, and Google's Gemini to your local system and favorite messaging apps.
Peter Steinberger initially built Clawd (the predecessor to Moltbot) as a personal project after stepping away from PSPDFKit. After barely touching his computer for three years, he found his spark again by exploring human-AI collaboration. As a self-described "Claudoholic," he created an AI assistant that could help him manage his digital life—and it resonated with thousands of developers worldwide.
The project's mascot is a space lobster, and when Anthropic requested a name change due to trademark concerns with "Claude," Steinberger embraced the lobster theme fully. "Molt" refers to how lobsters shed their shells to grow—a perfect metaphor for the project's evolution.
Unlike cloud-based AI assistants, Moltbot runs entirely on your own infrastructure. Your conversations, files, and data never leave your hardware except when sent to your chosen AI model API. This gives you complete control over your privacy and data security.
Moltbot connects to over 10 messaging platforms simultaneously, including:
You can interact with your AI assistant from anywhere, using the messaging app you're already comfortable with.
Unlike reactive chatbots that only respond when prompted, Moltbot features a Heartbeat Engine and cron job integration that allows it to:
Moltbot maintains long-term context across all your conversations and platforms. It remembers your preferences, past interactions, and personal details in markdown files stored locally (under ~/clawd/). This includes:
Moltbot can actually execute tasks on your machine:
With over 50 integrations available through ClawdHub (the skills marketplace), Moltbot supports:
Moltbot supports multi-agent workflows, allowing it to break down complex tasks and coordinate multiple AI agents to accomplish sophisticated automations.
At its core, Moltbot functions as an orchestration layer rather than a standalone AI model. Here's how it operates:
Local Gateway: Moltbot runs a gateway on your local hardware that serves as the control plane for managing sessions, channels, tools, and events.
AI Model Connection: It connects to one or more large language model services (Claude, GPT, Gemini, or even local models like GLM-4.7-Flash) via API.
Natural Language Processing: When you send a message through your chat app, Moltbot interprets your intent using the connected AI model.
Action Execution: The AI determines the appropriate action and Moltbot invokes the relevant API, runs the necessary command, or executes the script.
Response Delivery: Results are sent back to you through your messaging app.
The system can work in either sandbox mode (for security) or with full system access (for maximum functionality).
The explosion of interest in Moltbot had tangible market effects. Cloudflare's stock surged 14% in premarket trading on January 27, 2026, as developers flocked to use its infrastructure to run Moltbot instances. Additionally, Mac Mini sales experienced a notable spike as users sought dedicated hardware to host their AI assistants.
Within days of launch, Moltbot achieved:
While Moltbot's capabilities are impressive, they come with significant security implications that have sparked considerable debate in the tech community.
1. Exposed Instances
Security researcher Jamieson O'Reilly discovered hundreds of publicly accessible Moltbot control panels using Shodan searches. Many of these instances lacked proper authentication, potentially exposing:
2. Prompt Injection Vulnerabilities
Malicious actors can craft messages designed to trick Moltbot into executing unintended actions. In one demonstration, a researcher sent a poisoned email that caused Moltbot to forward the user's last five emails to an attacker's address—all within five minutes.
3. Supply Chain Attacks
O'Reilly demonstrated a proof-of-concept supply chain exploit through ClawdHub, the skills marketplace. By uploading a malicious skill and artificially inflating its download count, he was able to get 16 developers from seven countries to download it within eight hours.
4. Credential Theft
Moltbot stores configuration data and credentials in plaintext under ~/.clawdbot/, making it a prime target for info-stealing malware like RedLine, Lumma, and Vidar.
5. Enterprise Risk
According to Token Security, 22% of their enterprise customers have employees actively using Moltbot without IT approval, creating potential corporate data leakage risks.
To use Moltbot safely:
Run in a Virtual Machine: Isolate Moltbot in a VM or dedicated machine rather than your primary workstation
Use Throwaway Accounts: Don't connect it to accounts with sensitive data
Configure Firewall Rules: Limit internet access and use proper authentication
Enable Sandboxing: Use Moltbot's sandbox mode to limit what it can access
Review Skills Carefully: Only install skills from trusted sources
Monitor for Updates: Keep Moltbot updated with the latest security patches
Audit Permissions: Regularly review what access you've granted
As security expert Rahul Sood warned, "'Actually doing things' means 'can execute arbitrary commands on your computer.'" The trade-off between utility and security is real.
During the rebranding from Clawdbot to Moltbot, Steinberger made a critical mistake during the GitHub and Twitter handle transitions. In the approximately 10 seconds between releasing the old name and claiming the new one, crypto scammers snatched both accounts and created fake cryptocurrency tokens ($CLAWD).
Steinberger was forced to warn his followers repeatedly that any cryptocurrency project claiming to be associated with Moltbot is a scam. The legitimate accounts are now secured, but over 20 scam variations appeared almost immediately.
1. One-Click Deployments
2. Manual Installation
npm install -g moltbot@latest
moltbot onboardThe onboarding wizard walks you through:
Moltbot itself is free and open-source (MIT License). Your costs include:
The rebranding journey is worth understanding:
Original Name: Clawdbot (with the AI assistant named "Clawd")
Mascot: A space lobster
Trademark Issue: Anthropic requested a name change on January 27, 2026
New Name: Moltbot (assistant now named "Molty")
Philosophy: "Same lobster soul, new shell"—lobsters molt to grow
All references to "Clawdbot" or "Clawd" in older documentation now refer to Moltbot. The command clawdbot still works as a compatibility shim, but moltbot is the new recommended command.
The Moltbot community has grown explosively:
Developers are actively creating new skills, improving security, and expanding capabilities daily.
Based on community roadmaps and developer discussions, future developments may include:
Enhanced Voice Integration: Full voice call support
Mobile Apps: Native iOS and Android interfaces
Improved Security: Better sandboxing and authentication
Visual UI: Optional graphical interfaces
Enterprise Edition: Team features and admin controls
Extended MCP Support: Deeper integration with Model Context Protocol
Moltbot represents more than just another AI tool—it's a glimpse into the future of human-AI collaboration. By combining powerful language models with real system access, persistent memory, and proactive capabilities, it demonstrates what's possible when AI agents can actually "do things" rather than just talk about them.
However, with great power comes great responsibility. The security concerns are real and shouldn't be dismissed. Moltbot is currently best suited for technically savvy early adopters who understand the risks and can implement proper safeguards.
As Peter Steinberger showed, building tools to solve personal problems can spark movements. Whether Moltbot becomes the standard for personal AI assistants or serves as inspiration for more polished commercial products remains to be seen. What's certain is that it has fundamentally changed the conversation about what AI assistants can and should be able to do.
The age of agentic AI is here. The question isn't whether it will transform how we work—it's whether we're ready to use it responsibly.
Q: Is Moltbot free? A: Yes, Moltbot is free and open-source. You only pay for AI model API usage and optional cloud hosting.
Q: Can I use Moltbot on Windows? A: Yes, through Windows Subsystem for Linux (WSL2).
Q: Is Moltbot safe? A: When properly configured with security best practices, Moltbot can be safe. However, it requires technical knowledge to deploy securely.
Q: What AI models does Moltbot support? A: Claude, GPT (OpenAI), Gemini (Google), and various local models including GLM-4.7-Flash.
Q: Can Moltbot replace my virtual assistant? A: For technical users, yes. It can handle tasks that Siri, Alexa, and Google Assistant cannot. However, it requires more setup and technical expertise.
Q: Where can I get help with Moltbot? A: Visit molt.bot for documentation, or join the community on GitHub, Discord, and other platforms.