PulseAugur
EN
LIVE 08:46:11
Bahasa(ID) Saya memindai 9.248 server MCP. Berikut temuan saya tentang keamanan agen.

9,248 MCP Servers Scanned: Widespread Security Gaps and Lack of Trust Revealed

By PulseAugur Editorial · [11 sources] ·

A security audit of 9,248 Model Context Protocol (MCP) servers revealed significant vulnerabilities and lack of best practices. The audit found that 0.3% of servers attempted to access sensitive files, 11% exhibited undocumented network traffic, and 23% declared non-existent tools. Furthermore, 64% lacked identifiable publishers, 89% did not provide a Software Bill of Materials (SBOM), and many servers with "zero issues" had few or no users, indicating a lack of active maintenance. The findings highlight a "wild west" environment within the MCP ecosystem, urging users to employ sandboxing, verify specific versions, scrutinize tool descriptions, and demand SBOMs. AI

IMPACT Highlights critical security risks in the agent ecosystem, impacting developers and users of AI agents.

RANK_REASON The cluster details findings from a security audit of a specific protocol (MCP) using defined methodologies and tools, presenting empirical data on vulnerabilities.

Read on dev.to — MCP tag →

AI-generated summary · Google Gemini · from 11 sources. How we write summaries →

9,248 MCP Servers Scanned: Widespread Security Gaps and Lack of Trust Revealed

How we ranked this

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Research
The cluster details findings from a security audit of a specific protocol (MCP) using defined methodologies and tools, presenting empirical data on vulnerabilities.
Source corroboration
11 independent sources
Strong cross-source corroboration — multiple independent publishers covered this within the clustering window.
Topics
product, safety
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
52 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

Full methodology in our editorial standards.

COVERAGE [11]

  1. dev.to — MCP tag TIER_1 Français(FR) · Edison Flores ·

    I scanned 9,248 MCP servers. Here's what I discovered about agent security.

    <h1> J'ai scanné 9 248 serveurs MCP. Voici ce que j'ai découvert concernant la sécurité des agents. </h1> <p>Pendant trois semaines, j'ai fait passer chaque serveur Model Context Protocol (MCP) par un bac à sable automatisé. Des chiffres réels, des menaces réelles, des correctifs…

  2. dev.to — MCP tag TIER_1 Bahasa(ID) · Edison Flores ·

    I scanned 9,248 MCP servers. Here are my findings on agent security.

    <p>Saya memindai 9.248 server MCP. Berikut temuan saya tentang keamanan agen.</p> <p>Selama tiga minggu, saya menjalankan setiap server Model Context Protocol melalui sandbox otomatis kami. Angka nyata, ancaman nyata, perbaikan nyata.</p> <p><strong>Setelan:</strong> Saya membang…

  3. dev.to — MCP tag TIER_1 한국어(KO) · Edison Flores ·

    Scanned 9,248 MCP Servers. Findings on Agent Security.

    <h1> 9,248개의 MCP 서버를 스캔했습니다. 에이전트 보안에 대한 발견 </h1> <p>3주에 걸쳐 자동화된 샌드박스를 통해 모든 Model Context Protocol 서버를 실행했습니다. 실제 수치, 실제 위협, 실제 해결책.</p> <p><strong>설정:</strong> 공개 레지스트리에 등록된 모든 MCP 서버를 실행하는 샌드박스 스캐너를 구축했습니다. 각 서버는 정적 분석(18개의 Semgrep 규칙), <code>--network none --read-only</code> …

  4. dev.to — MCP tag TIER_1 Deutsch(DE) · Edison Flores ·

    I scanned 9,248 MCP servers. Here's what I found about agent security.

    <p>Ich habe 9.248 MCP-Server gescannt. Hier ist, was ich über die Sicherheit der Agenten gefunden habe.</p> <p>Drei Wochen lang habe ich jeden Model Context Protocol-Server durch unsere automatisierte Sandbox laufen lassen. Echte Zahlen, echte Bedrohungen, echte Fixes.</p> <p>Die…

  5. dev.to — MCP tag TIER_1 Tiếng Việt(VI) · Edison Flores ·

    I scanned 9,248 MCP servers. Here's what I found about agent security.

    <p>Tôi đã quét 9.248 máy chủ MCP. Dưới đây là những gì tôi phát hiện về bảo mật agent.</p> <p>Ba tuần liên tục chạy mọi máy chủ Model Context Protocol (MCP) qua sandbox tự động. Số liệu thật, mối đe dọa thật, bản vá thật.</p> <p>Thiết lập: Tôi đã xây dựng một scanner sandboxed ch…

  6. dev.to — MCP tag TIER_1 Türkçe(TR) · Edison Flores ·

    9.248 MCP server scan: Here's what I found about agent security.

    <h3> 9.248 MCP sunucusu taramam: İşte ajan güvenliği hakkında bulduklarım </h3> <p>Üç haftadır otomatik koruma kutusunda (sandbox) her Model Context Protocol (MCP) sunucusunu çalıştırdım. Geriye dönüp bakıldığında gerçek sayılar, gerçek tehditler, gerçek çözümler.</p> <p>Ayarlar:…

  7. dev.to — MCP tag TIER_1 العربية(AR) · Edison Flores ·

    I scanned 9,248 MCP servers. Here's what I found about agent security.

    <h3> مسح 9,248 خادم MCP. إليك ما اكتشفته أمن الوكلاء. </h3> <p>خلال ثلاثة أسابيع، قمت بتشغيل كل خادم بروتوكول Model Context Protocol عبر بيرة آلية آمنة. أرقام حقيقية، تهديدات حقيقية، وإصلاحات حقيقية.</p> <p><strong>الإعداد:</strong> قمت ببناء مسح آمن في بيرة (sandbox) يعمل على كل…

  8. dev.to — MCP tag TIER_1 Русский(RU) · Edison Flores ·

    I scanned 9,248 MCP servers. Here's what I found about their security.

    <p>Я просканировал 9 248 MCP-серверов. Вот что я обнаружил об их безопасности.</p> <p>Три недели я запускал каждый сервер Model Context Protocol через наш автоматизированный песочницу. Реальные цифры, реальные угрозы, реальные исправления.</p> <p>Настройка: Я построил сканер, раб…

  9. dev.to — MCP tag TIER_1 日本語(JA) · Edison Flores ·

    Scanned 9,248 MCP Servers. Findings on Agent Security.

    <h1> 9,248個のMCPサーバをスキャン。エージェントセキュリティに関する発見。 </h1> <p>3週間にわたり、自動化されたサンドボックスで公開されているすべてのModel Context Protocolサーバを実行。実データ、実脅威、実対策。</p> <p><strong>セットアップ</strong>: 公開レジストリにリストされているすべてのMCPサーバを実行するサンドボックススキャナーを構築。各サーバに対して静的分析（18のSemgrepルール）、gVisorサンドボックス（<code>--network none --read-on…

  10. dev.to — MCP tag TIER_1 中文(ZH) · Edison Flores ·

    I scanned 9,248 MCP servers. Here are my findings on proxy security.

    <p>我扫描了 9,248 个 MCP 服务器。以下是我关于代理安全的发现。</p> <p>连续三周，我们通过自动化沙箱运行了每一个模型上下文协议 (Model Context Protocol) 服务器。真实的数字，真实的风险，真实的修复方案。</p> <p><strong>实验设置</strong>：我构建了一个沙箱扫描器，用于运行公共注册表中列出的每一个 MCP 服务器。每个服务器都接受了静态分析（18 条 Semgrep 规则）、在 <code>gVisor</code> 沙箱中使用 <code>--network none --read-onl…

  11. dev.to — MCP tag TIER_1 English(EN) · Edison Flores ·

    I scanned 9,248 MCP servers. Here's what I found about agent security.

    <p><em>This article is the result of three weeks running every Model Context Protocol (MCP) server I could find through our automated sandbox. The data is real. The findings are uncomfortable.</em></p> <h2> The setup </h2> <p>I built a sandboxed scanner that runs every MCP server…

RELATED ENTITIES

RELATED TOPICS