OpenClaw hits 100K GitHub stars after third rebrand, 2M weekly visits
What's the deal? The viral AI assistant that couldn't keep a name has settled on its third identity in less than a week. OpenClaw, formerly Moltbot and originally Clawdbot, has surpassed 100,000 GitHub stars and attracted over 2 million website visits in seven days. Created by Austrian developer Peter Steinberger, the open-source agent runs locally on users' devices and integrates with WhatsApp, Slack, Telegram and other messaging platforms to manage emails, calendars and system commands autonomously.
This rebrand appears more permanent. Steinberger conducted trademark searches before launch, secured domains and prepared migration code — a stark contrast to the previous hasty rename that triggered account hijackings and a fake cryptocurrency scam.
The name combines "Open" to emphasise community-driven, self-hosted development with "Claw" to retain the project's lobster mascot. The team states this name "will stay."
Why now? OpenClaw represents a pivot from viral chaos to infrastructure maturity. The project has shipped 34 security-related commits, released machine-readable security models and repositioned itself as model-agnostic infrastructure rather than a Claude wrapper.
New integrations include KIMI K2.5 and Xiaomi MiMo-V2-Flash models, plus Twitch and Google Chat plugins. The team is recruiting maintainers and establishing formal processes for handling pull requests and issues.
IBM researchers featured OpenClaw on their Mixture of Experts podcast, with principal research scientist Kaoutar El Maghraoui noting it challenges assumptions that autonomous AI agents require vertical integration. "Creating agents with true autonomy and real-world usefulness is not limited to large enterprises," she said. "It can also be community driven."
What could go wrong? Security remains the primary concern. El Maghraoui warned that highly capable agents without proper safety controls create major vulnerabilities, particularly in work contexts. The project acknowledges prompt injection remains an unsolved industry problem.
The loose, open-source architecture that makes OpenClaw powerful also makes it dangerous if misconfigured. Earlier versions exposed hundreds of instances to the public internet, potentially leaking months of private messages, credentials and API keys.
The sudden leadership transition risk persists. With a single developer at the helm and growing infrastructure demands, sustainability questions remain despite plans to compensate full-time maintainers.
The signal: OpenClaw demonstrates that useful AI agents needn't be built by tech giants. The project's trajectory from weekend hack to 100,000-star repository shows demand for AI tools that actually execute tasks rather than just suggest them.
More significantly, it forces a conversation about integration architecture. El Maghraoui predicts successful agents will adopt "hybrid integration" — modular platforms that integrate deeply when security demands it but remain flexible enough to run locally across domains.
The shift from "cool hack" to "serious infrastructure" energy signals maturity in the agentic AI space. What comes next will likely balance OpenClaw's community-driven flexibility with the security rigor currently reserved for enterprise systems.
Sources:
Dev.to
Business Today
AI Base
IBM Think
Cnet
B.S.
