OpenClaw: 270K Exposed Amid Uninstall Surge

🔑 Enhanced Key Takeaways

• •OpenClaw, formerly known as Moltbot and Clawdbot, achieved over 150,000 GitHub stars shortly after release due to its viral adoption for autonomous AI agent capabilities[4].
• •CVE-2026-25253 (CVSS 8.8) enables remote code execution via malicious JavaScript that leaks gateway authentication tokens, patched in version 2026.1.29[1][5].
• •Infostealers like RedLine, Lumma, and Vidar target OpenClaw's plaintext config files storing API keys and OAuth tokens[1].
• •ClawHub skill marketplace has faced repeated compromises with malicious skills, exacerbating risks beyond default configurations[1].

🛠️ Technical Deep Dive

• •OpenClaw gateway ships with authentication disabled by default, exposing instances to public internet access without firewalls[1].
• •Supports 'skills' as plugins for actions like file access, web browsing, and command execution, with autonomous decision-making on skill chaining[2].
• •Vulnerable to prompt injection attacks where malicious instructions in webpages, messages, or PDFs override programmed behavior[2][3].
• •Credentials including API keys, OAuth tokens, and bot credentials stored in plaintext config files[1].

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline