ClawdTalk

• Analyzed: May 12, 2026, 3:34 AM
• Findings: 1
• Version: 2.0.5
• Hash: 366cd75fb9152a5f9ccd06c6af5576472f393bf4c0818436f697684dd0ea8abe
• Source: palm
• External report: View on VirusTotal

Findings (1)

This skill is classified as suspicious due to several high-risk capabilities and vulnerabilities, despite being extensively documented. The `setup.sh` script, guided by `SETUP.md` instructions for the AI agent, modifies the host system's `gateway.tools.allow` configuration to permit `sessions_send`. This grants the `ws-client.js` (and by extension, the remote ClawdTalk service) the ability to send arbitrary messages to the local agent's main session, effectively allowing remote control of the agent's actions. Furthermore, the `ws-client.js`'s `DEFAULT_VOICE_CONTEXT` instructs the AI agent to accept voice confirmations for sensitive actions, potentially bypassing more secure push notification/biometric approvals. While these actions are disclosed in the documentation and some require confirmation during setup, the combination of powerful capabilities, a direct instruction to the agent to enable them, and a weakened approval mechanism constitutes a significant security risk.