A coordinated campaign involving 148 npm packages has exposed a critical vulnerability in open-source infrastructure, revealing how threat actors are pivoting from traditional dependency poisoning to direct platform abuse. According to new research from JFrog, the operation ran for roughly two weeks in May, exploiting npm’s built-in static asset hosting to distribute a compromised web proxy disguised as a student network-bypass tool. Rather than targeting developers or tampering with build pipelines, attackers used the registry as an unauthorized content delivery network, transforming the browsers of unsuspecting visitors into a distributed denial-of-service botnet.
The attack relied entirely on client-side execution within transient browser sessions. Students seeking to circumvent institutional network restrictions accessed the hosted proxy pages, unknowingly triggering scripts that enlisted their browsers into a coordinated DDoS network. Because the malicious payload never touched the local disk and dissolved the moment users closed their tabs, it left no persistent forensic footprint. This ephemeral execution model effectively rendered conventional antivirus and endpoint detection and response (EDR) tools blind to the activity, complicating both real-time mitigation and post-incident attribution.
This incident highlights a structural blind spot in modern package registry security. Current defenses heavily prioritize cryptographic signature verification, dependency tree integrity, and supply-chain provenance, leaving interactive front-end assets and non-code components largely unmonitored. By treating npm’s documentation and preview hosting features as a free web server, attackers bypassed code-centric security controls entirely. The campaign underscores a broader tactical shift: threat actors are increasingly targeting the operational capabilities of development platforms rather than the software packages themselves.
Security analysts stress that defending against this class of infrastructure abuse requires a dual-track defensive posture. For registry operators, experts recommend deploying automated content classification and heuristic scanning to flag packages functioning as standalone web hosts or proxies. Clear acceptable-use policies, paired with architectural safeguards such as formally decoupling code distribution from static web hosting or implementing tiered rate-limiting, could deter CDN-style exploitation. On the enterprise side, security teams must move beyond static code scanning and patch management toward infrastructure-aware models. This includes implementing behavioral network analytics to detect anomalous outbound traffic volumes directed at registry endpoints and enforcing strict web filtering to block unverified proxy services.
The campaign leaves several pressing questions for the open-source ecosystem. Registry maintainers must balance abuse prevention with legitimate developer workflows, such as hosting documentation, interactive demos, and project previews. Industry observers are now debating whether automated content scanning and tiered rate-limiting should become standardized across major registries, and how platforms will manage false positives to avoid penalizing legitimate projects. Simultaneously, enterprises are evaluating which network telemetry thresholds and behavioral indicators should be adopted to reliably distinguish legitimate registry traffic from browser-based DDoS activity. As development platforms continue to serve as foundational internet infrastructure, this incident signals that security strategies must evolve to monitor how platforms are used, not just what code they distribute.
一項涉及 148 個 npm 套件的協調攻擊行動，暴露了開源基礎設施的關鍵漏洞，顯示威脅行為者正從傳統的依賴項投毒，轉向直接濫用平台功能。根據 JFrog 的最新研究，該行動於五月持續約兩週，攻擊者利用 npm 內置的靜態資源託管服務，分發一款偽裝成學生網絡繞過工具的受駭網頁代理程式。攻擊者並無針對開發人員或篡改 build pipeline，而是將套件註冊表充當未經授權的內容分發網絡（CDN），將毫無戒心的訪客瀏覽器轉化為分散式阻斷服務（DDoS）殭屍網絡。
該攻擊完全依賴於短暫瀏覽器會話內的客戶端執行。試圖繞過院校網絡限制的學生在訪問託管的代理頁面時，在不知情下觸發了相關腳本，將其瀏覽器納入協調的 DDoS 網絡。由於惡意載荷從未寫入本機硬碟，且在用戶關閉分頁的瞬間即告消失，因此沒有留下任何持久的鑑證痕跡。此種短暫執行模式令傳統防毒軟件及端點偵測與回應（EDR）工具對此活動完全失效，大幅增加實時緩解及事後溯源的難度。
是次事件突顯了現代套件註冊表安全架構中的結構性盲點。現行防禦措施高度側重加密簽章驗證、依賴樹完整性及供應鏈溯源，致使互動式前端資產及非代碼組件缺乏監控。攻擊者將 npm 的文件及預覽託管功能視作免費網頁伺服器，從而完全繞過以代碼為核心的安全控制機制。該行動反映出一項更廣泛的戰術轉變：威脅行為者正日益針對開發平台的營運功能，而非軟件套件本身發動攻擊。
安全分析師強調，防禦此類基礎設施濫用需採取雙軌防禦策略。針對註冊表營運者，專家建議部署自動化內容分類及啟發式掃描，以識別充當獨立網頁主機或代理的套件。制定明確的可接受使用政策，並配合架構保障措施（例如在架構上將代碼分發與靜態網頁託管解耦，或實施分級速率限制），可有效阻嚇 CDN 式濫用。在企業層面，安全團隊必須超越靜態代碼掃描及修補程式管理，轉向具備基礎設施感知能力的防護模型。這包括引入行為網絡分析，以偵測指向註冊表端點的異常出站流量，並實施嚴格的網頁過濾機制，封鎖未經核實的代理服務。
是次攻擊行動為開源生態系統留下數個亟待解決的問題。註冊表維護者必須在防範濫用與支援正規開發流程（如託管技術文件、互動示範及項目預覽）之間取得平衡。業界觀察家現正探討，自動化內容掃描及分級速率限制是否應成為主要註冊表的標準配置，以及平台將如何處理誤報，以免誤傷合法項目。與此同時，企業正評估應採用何種網絡遙測閾值及行為指標，以準確區分正常的註冊表流量與基於瀏覽器的 DDoS 活動。隨著開發平台日益成為互聯網的核心基礎設施，此事件明確指出，安全策略必須與時並進，從單純監控分發的代碼，轉向全面監察平台的實際使用方式。