baoyu-post-to-x

The skill is internally consistent with its stated purpose, but it is high risk because it empowers an AI agent to publish publicly to X using the user's live browser session and anti-bot bypass automation. This looks more like a legitimate but sensitive automation skill than malware; the primary concern is autonomous real-world posting, not covert exfiltration.

No clear evidence of deliberate malware/backdoor behavior in this module. However, it is security-sensitive: it downloads arbitrary attacker-controlled HTTP(S) URLs (including redirects) and writes the responses to disk, creating an SSRF/uncontrolled egress risk. Separately, its Markdown-to-HTML conversion performs insufficient escaping and inserts unvalidated link href values directly into HTML, creating a high XSS/HTML-injection risk if the resulting HTML is rendered or served without robust sanitization at a higher layer.

This module is a dual-use cross-platform keystroke injection utility that performs a real paste (Cmd+V/Ctrl+V equivalents) to the active/frontmost application and explicitly aims to bypass synthetic paste-event detection. No network exfiltration or persistence is present in the provided code, but the behavior is inherently security-sensitive (UI input automation) and can be abused for clipboard-driven attacks. The highest concrete technical risk is on macOS: user-supplied --app is interpolated into dynamically constructed AppleScript executed via osascript without escaping/validation, creating an AppleScript-evaluation injection risk. Overall: elevated security risk; treat as high-impact automation code and restrict/validate inputs when used.