惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Apple Machine Learning Research
Apple Machine Learning Research
爱范儿
爱范儿
博客园_首页
博客园 - 【当耐特】
V
Visual Studio Blog
博客园 - 叶小钗
月光博客
月光博客
美团技术团队
J
Java Code Geeks
小众软件
小众软件
Y
Y Combinator Blog
博客园 - Franky
Martin Fowler
Martin Fowler
博客园 - 聂微东
Microsoft Azure Blog
Microsoft Azure Blog
IT之家
IT之家
MyScale Blog
MyScale Blog
人人都是产品经理
人人都是产品经理
Microsoft Security Blog
Microsoft Security Blog
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
阮一峰的网络日志
阮一峰的网络日志
酷 壳 – CoolShell
酷 壳 – CoolShell
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
云风的 BLOG
云风的 BLOG

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
One npm Account Publishes 964 Million Downloads Per Week....
Pico · 2026-06-18 · via DEV Community

Pico

The npm account ai publishes seven packages. Combined, they install 964 million times per week:

Package Weekly downloads Publishers Risk
postcss 245,612,332 1 CRITICAL
nanoid 206,588,788 1 CRITICAL
caniuse-lite 173,435,668 1 CRITICAL
browserslist 167,746,012 1 CRITICAL
autoprefixer 63,517,741 1 CRITICAL
postcss-nested 54,486,292 1 CRITICAL
postcss-js 52,771,544 1 CRITICAL

That's 50 billion installs per year behind a single set of npm credentials. None of them have npm provenance attestations.

Why this matters

npm provenance uses OIDC tokens from GitHub Actions instead of long-lived npm tokens. If a package has provenance, you can verify that the published code came from a specific commit in a specific repository — not from someone's compromised laptop.

Without provenance, there's no way to distinguish a legitimate release from one pushed by a stolen token. The blast radius here is nearly a billion installs per week.

This isn't theoretical. axios was attacked on March 30, 2026 through a stolen npm token — same single-publisher, no-provenance pattern. LiteLLM was hit the same way a month earlier. The Shai-Hulud worm in May 2026 exploited stolen tokens to republish 637 package versions in 39 minutes.

What makes this different from chalk or lodash

PostCSS is interesting because it's not just one critical package. It's an entire ecosystem of critical packages, all behind the same account. chalk is one package, one publisher, 432M downloads/week. Bad enough. But ai controls seven independent packages that each cross the 10M threshold.

A compromised ai token doesn't just hit postcss. It hits the CSS build pipeline (postcss + autoprefixer + postcss-nested + postcss-js), the browser compatibility layer (browserslist + caniuse-lite), and one of the most popular ID generators in the ecosystem (nanoid).

And caniuse-lite was flagged with a dormant publisher warning — 61 months of inactivity on the publishing account. postcss-nested hasn't had a release in over 12 months.

This has been fixed before

fast-xml-parser (88M downloads/week, single publisher) had the same problem. After the community raised the issue, the maintainer set up GitHub Actions OIDC publishing. Within days, version 5.9.1 shipped with SLSA provenance attestations. Then 5.9.2 added environment gates and SHA-pinned actions. The structural gap closed in under a week.

I filed an issue on PostCSS yesterday proposing the same approach. The fix is a one-line change — add provenance: true to the npm publish step — and it requires zero stored secrets.

Check your own dependencies

If you want to see which packages in your project have this concentration risk:

npx proof-of-commitment

Run it in any project directory. It auto-detects your lockfile and flags packages where a single npm publisher controls more than 10M weekly downloads. That's the exact attack surface that's been exploited three times in four months.

The full PostCSS ecosystem audit data comes from Commit, which scores packages on behavioral signals rather than declared metadata.