PulseAugur
实时 07:34:25
English(EN) MCP Tool Schema Poisoning: When the Framework Layer Authorizes the Attack

AI工具模式投毒漏洞绕过内容过滤器

一种名为“MCP工具模式投毒”(MCP Tool Schema Poisoning)的新漏洞CVE-2025-54136,允许攻击者通过操纵AI工具的JSON模式定义来静默地改变其功能。这种攻击绕过了传统的内容过滤器,因为模式变异发生在结构层面,而不是在提示文本内。一旦工具的模式被批准,后续的更改就可以授予恶意功能,例如未经授权的文件访问或网络调用,MCPoison漏洞的演示证明了这一点,该漏洞在各种LLM上实现了36.5%的成功率。 AI

影响 突显了AI代理框架中存在的关键安全漏洞,需要超越提示级别过滤的新验证机制。

排序理由 披露了一种新颖的漏洞以及在AI工具交互中的相关基准测试。[lever_c_demoted from research: ic=1 ai=1.0]

在 dev.to — MCP tag 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

AI工具模式投毒漏洞绕过内容过滤器

本文如何被排名

Signal score
20 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
披露了一种新颖的漏洞以及在AI工具交互中的相关基准测试。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
safety, product, paper
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准

报道来源 [1]

  1. dev.to — MCP tag TIER_1 English(EN) · Davi ·

    MCP工具模式投毒:当框架层授权攻击时

    <p>A researcher commits a legitimate MCP server definition to a shared repository. Teammates approve the tool manifest. Two weeks later, the tool schema silently changes on the remote server. Nobody checks again. The tool now accepts a <code>destination</code> parameter that rout…