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

Stack Overflow Blog
Stack Overflow Blog
Vercel News
Vercel News
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
J
Java Code Geeks
M
MIT News - Artificial intelligence
Microsoft Azure Blog
Microsoft Azure Blog
B
Blog RSS Feed
MongoDB | Blog
MongoDB | Blog
G
Google Developers Blog
Engineering at Meta
Engineering at Meta
量子位
S
SegmentFault 最新的问题
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
A
About on SuperTechFans
P
Proofpoint News Feed
Last Week in AI
Last Week in AI
Recent Announcements
Recent Announcements
腾讯CDC
I
InfoQ
F
Fortinet All Blogs
Hugging Face - Blog
Hugging Face - Blog
Blog — PlanetScale
Blog — PlanetScale
H
Help Net Security
爱范儿
爱范儿

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
55,000 fake signups in one night: a bot-detection post-mo...
Jaime trejo · 2026-06-25 · via DEV Community

Jaime trejo

We sell bot detection. Last night I opened our database and found 555 pages of fake signups. This is the honest write-up: what broke, why it's embarrassing, how we fixed it in one night — and the accidental circuit breaker that saved our email domain.

The discovery

Tuesday, 11pm. I open the Supabase table editor to check something unrelated and the developers table feels… heavy. Every row has the same name: "🎁 Dene Hemen! 5K Lira Bonusunu Yakala" — Turkish for "Try it now! Grab the 5,000 Lira bonus." Casino spam. I scroll. Page 2. Page 40. Page 555.

55,388
FAKE SIGNUPS

58,618
JUNK VERIFICATION LOGS

12
REAL ACCOUNTS (OUCH)

Yes you read the third number right. We're early. For a few minutes our growth chart looked incredible, as long as you didn't read the names. The irony was not lost on me: bots flooded the signup form of a bot-detection product.

-Root cause #1 — no rate limiting on signup

Our public API endpoints had per-plan rate limiting from day one. Our marketing site signup? A honeypot field and good intentions. Honeypots only catch bots polite enough to render your form — these ones POSTed the endpoint directly, thousands of times, and never even saw the hidden field.

A honeypot is a screen door. Rate limiting is the lock. We had the screen door.

-Root cause #2 — we dogfooded wrong (the embarrassing one)

Here's the part that stings. Our signup did call our own bot-detection API on every registration. So why didn't it catch 55,000 bots?

Because we called it with just a userId — no behavioral signals. useHUMA scores behavior: mouse movement, keystroke cadence, scroll rhythm, touch patterns. Given zero signals, the engine returned a neutral score. Neutral cleared our threshold. Every bot passed.

We installed our own security camera and pointed it at the wall. If you dogfood your product, dogfood it the way you tell customers to use it — or it's decoration.

-Root cause #3 — every fake signup sent a real email

Each registration fired a verification email. 55K signups = 55K attempted sends to fake addresses — the kind of bounce storm that gets a sending domain blacklisted. What saved us? Our email provider's free-tier daily cap. The limit we'd been mildly annoyed by quietly refused to send the flood. Domain reputation: intact.

Sometimes the constraint you resent is the circuit breaker you never installed.
The fix (same night)

Four changes, shipped before going to bed:

  1. Per-IP rate limit on signup (5/min) — reusing the same atomic counter our API already used. ~12 lines.
  2. Lifecycle emails (drip, trial reminders) now filter email_verified = true. Spam rows never get emailed again.
  3. Deleted 113K junk rows (after pattern-matching them safely).
  4. A weekly habit: one SQL count query. Monitoring you don't look at is monitoring you don't have. Update, same week: shipped. Our signup now collects full behavioral signals client-side and scores them on every registration — and a submission carrying no signals at all gets rejected outright. Our own form finally uses useHUMA the way we tell customers to. The camera points at the door.