thai-contextual-editor

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill explicitly instructs the agent to "preserve code, URL, keys, ... placeholders verbatim" so if the user input contains secrets or API keys the agent is required to reproduce them in its output (even though a later line warns not to send credentials to workers), which creates an exfiltration risk.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). SKILL.md specifies that the agent ingests user-provided text as the “Entry”/request content to be analyzed and rewritten (e.g., “มีร่างภาษาไทย…หรือ brief” in Phase 1, and “แก้เฉพาะไฟล์หรือข้อความที่ผู้ใช้ระบุ”), so outsider-authored free text is directly read at runtime as the source/brief.