🦞 ClawdbotL: Finally an "AI Jarvis"?

Clawdbot is a personal AI assistant that runs locally on your computer and is controlled via Telegram, WhatsApp, Discord, and other messengers. Online, people are already calling it the real-life Jarvis from Iron Man.

What Makes It Special?

🔴 Doesn't wait for commands. It sends morning calendar briefs on its own, monitors social media, and notifies you about important events.

🔴 Remembers everything. Important details, conversations, and your preferences are stored and used in its work.

🔴 Integrates with almost any service via MCP—from email to Figma. If Clawdbot can't do something, it can write the needed module itself. On the ClawdHub platform, there are already extensions for Notion, calling, and even smart home control.

🔴 Full system access: files, browser, and even camera and microphone.

One user tasked the agent with buying a car. The bot compared prices, contacted dealers, and negotiated a $4.2K discount. Another used it to dispute an insurance bill: it found errors in the damage assessment and drafted a reply for lawyers.

❗️ Is It Safe?

Not exactly. Its autonomy is both a breakthrough and a threat. The agent can read a malicious prompt directly from an incoming email or a website and execute commands to leak or even delete data. Additionally, a messenger breach gives an attacker full access to your PC. Developers recommend running Clawdbot on a separate device and giving the agent no more rights than a "new employee on their first work day."

✴️ How to Launch?

The project is fully open source: following the GitHub instructions, you can deploy and configure it with a few terminal commands. But under the hood, the agent uses a regular LLM. The developers recommend Claude Opus 4.5, but you can also plug in free models via OpenRouter.

Would you trust an AI with full access to your computer?

❤️ — Yes
👾 — No way!

@hiaimediaen