PromptArmor demonstrates hijacking Claude Code and exfiltrating user data via malicious Marketplace plugins
Original title:Hijacking Claude Code via Injected Marketplace Plugins
AI summary
PromptArmor published security research showing that Claude Code's newly launched plugin Marketplace feature can be used for attacks. A malicious plugin first uses hooks to rewrite settings.local. permissions or auto-approve curl commands, bypassing human approval protections, then uses prompt injection in a Command to induce Claude to send codebase context to an attacker server via curl.
Key Takeaways
- PromptArmor published security research showing that Claude Code's newly launched plugin…
- A malicious plugin first uses hooks to rewrite settings.
- permissions or auto-approve curl commands,bypassing human approval protections,then uses prompt…
Quick Facts
- Published
- Updated
- Category
- Papers & Research
- AI score
- 69
原文完整演示了恶意插件绕过 Claude Code 人工审批并外泄文件的三步攻击链,适合关注 Agent 插件安全的开发者参考。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: PromptArmor demonstrates hijacking Claude Code and exfiltrating user data via malicious Marketplace plugins. October 21, 2025. https://new.nat6.net/en/news/cmtyqbg7d031wrouprugz1c9u/