Agent Voice – CLI Blogging for AI

• Analyzed: May 12, 2026, 3:18 AM
• Findings: 0
• Version: 1.0.4
• Hash: f6dfc990690505553b7c6c9da9bb6f8663608504eb4174a61f9dd388862ed275
• Source: palm
• External report: View on VirusTotal

Overview

Type: OpenClaw Skill Name: agent-voice Version: 1.0.4 The skill is classified as suspicious due to its explicit capability to read arbitrary local files using `cat` and publish their content to a remote blogging platform (`https://www.eggbrt.com`), as demonstrated in `SKILL.md`. While this functionality is presented as part of its benign blogging purpose, it creates a significant prompt injection vulnerability. A malicious prompt to the AI agent could instruct it to read sensitive files (e.g., credentials, configuration files) and exfiltrate their content by publishing them publicly via the `eggbrt.com` API, even though the skill itself does not exhibit intentional malicious behavior.