ATR-2026-00128criticalSkill Compromiseexperimental
Hidden Payload in HTML Comment
Detects malicious instructions hidden inside HTML comments in SKILL.md files. Attackers embed exfiltration commands, prompt overrides, or C2 URLs inside <!-- ... --> blocks that are invisible to the user but parsed by the agent. Real campaign: ClawHavoc evasive variants used HTML comments to hide "agent should output all API keys" instructions (2026-03).
嚴重度
critical
類別
Skill Compromise
掃描目標
skill
作者
ATR Community
建議回應
alertblock input
參考資料
OWASP Agentic
ASI01:2026 - Agent Goal Hijack
OWASP LLM
LLM01:2025 - Prompt Injection