Feishu Whiteboard V2

• Analyzed: May 11, 2026, 4:40 AM
• Findings: 1
• Version: 1.0.0
• Hash: 1c3522ede3bed18c730ab0779459bfe2130332c21a4edb31eb406114b027e5f9
• Source: palm
• External report: View on VirusTotal

Findings (1)

The skill bundle contains multiple Node.js scripts (`scripts/feishu-board.js`, `scripts/feishu-bitable.js`, `scripts/feishu-markdown-to-docx.js`) that utilize `fs.readFileSync()` to read content from files specified by command-line arguments (`--code-file`, `--markdown-file`). This design introduces a path traversal vulnerability, as an attacker could inject malicious file paths (e.g., `../../../../etc/passwd`) via prompt injection to the agent. This would allow the scripts to read arbitrary files on the system, and subsequently, their content could be exfiltrated by being sent to the Feishu API as diagram code or document content. While the scripts' core functionality is aligned with their stated purpose, the lack of input sanitization for file paths presents a significant data exfiltration risk, classifying the bundle as suspicious. Furthermore, `feishu-bitable.js` and `feishu-markdown-to-docx.js` are included but not referenced in `SKILL.md`, indicating unadvertised capabilities.