Castreader Openclaw Skill

• Analyzed: May 11, 2026, 6:18 AM
• Findings: 1
• Version: 2.1.1
• Hash: e0b2de7ea3a60933a030161ac305e83739949b3db7153189ec1f67d50e7db73b
• Source: palm
• External report: View on VirusTotal

Findings (1)

The skill bundle contains several high-risk security patterns and vulnerabilities. Most notably, it uses an unencrypted HTTP endpoint (http://api.castreader.ai:8123) to transmit web content extracted from sensitive platforms like Notion, Google Docs, and private AI chat interfaces (ChatGPT/Claude), exposing user data to MITM attacks. Additionally, SKILL.md provides instructions that create a shell injection risk by directing the agent to use 'echo' with unsanitized summary text. The 'extractor-bundle.js' is a large, complex payload designed to be injected into various high-privilege web sessions via Puppeteer, which, combined with the lack of transport security, poses a significant risk to data privacy.