OpenClaw AI Framework Hit by Malicious Extensions

OpenClaw, an open-source AI agent framework that enables autonomous task execution on users' local systems, has surged in popularity among developers but is grappling with escalating security threats, including malicious extensions and a recently patched critical vulnerability.

Launched initially as Clawdbot and renamed twice in quick succession to Moltbot and then OpenClaw amid trademark concerns, the tool operates as a self-hosted assistant capable of running terminal commands, managing files, and integrating with messaging platforms like WhatsApp and Telegram. Unlike traditional chatbots, OpenClaw maintains stateful interactions, remembers context, and performs actions such as API calls, workflow triggers, and data processing without relying on cloud services. This local execution appeals to technical users seeking data control and cost savings, particularly in regions like India where closed AI platforms dominate.

The framework's ecosystem, including ClawHub—a public registry for third-party "skills" or extensions—has become a vector for attacks. A security audit by Koi Security examined 2,857 skills and identified 341 malicious ones linked to multiple campaigns, with 335...