I Built a LinkedIn Spam Filter in a Weekend (And You Can Too)

Kevin Tran — Fri, 30 Jan 2026 01:09:37 +0000

I Built a LinkedIn Spam Filter in a Weekend (And You Can Too)

The Problem

LinkedIn has become unusable for job seekers.

Last month, I received 47 connection requests. 19 were "life coaches." 8 were forex traders. 6 were MLM recruiters. Only 7 were actual talent acquisition professionals.

I missed 2 legitimate recruiter messages because they were buried under spam.

The Solution

I built Request Radar - a Chrome extension that automatically labels LinkedIn invitations:

🟢 Recruiter - Don't miss these
🔴 Spam - Avoid these
🔵 Normal - Consider these

How It Works

1. Content Script Injection

// manifest.json
{
  "content_scripts": [{
    "matches": ["https://www.linkedin.com/*"],
    "js": ["content.js"],
    "run_at": "document_idle"
  }]
}

The extension injects a content script into LinkedIn pages. We use document_idle to ensure LinkedIn's content has loaded.

2. Detecting New Invitations

LinkedIn loads invitations dynamically. We use MutationObserver to watch for new content:

const observer = new MutationObserver(() => {
  scanInvitations();
});

observer.observe(document.body, { 
  childList: true, 
  subtree: true 
});

3. Classification Logic

Simple keyword matching:

function classifyInvitation(headline) {
  const recruiterKeywords = [
    'recruiter', 'talent acquisition', 
    'hiring manager', 'headhunter'
  ];

  const spamKeywords = [
    'life coach', 'passive income',
    'financial freedom', 'dm me'
  ];

  let recruiterScore = 0;
  let spamScore = 0;

  const text = headline.toLowerCase();

  recruiterKeywords.forEach(keyword => {
    if (text.includes(keyword)) recruiterScore++;
  });

  spamKeywords.forEach(keyword => {
    if (text.includes(keyword)) spamScore++;
  });

  if (recruiterScore > spamScore && recruiterScore > 0) {
    return 'recruiter';
  } else if (spamScore > recruiterScore && spamScore > 0) {
    return 'spam';
  }

  return 'normal';
}

4. Adding Visual Badges

function addBadge(card, type) {
  const badge = document.createElement('div');
  badge.className = 'radar-badge';

  const config = {
    recruiter: { emoji: '🟢', label: 'Recruiter', color: '#28a745' },
    spam: { emoji: '🔴', label: 'Spam', color: '#dc3545' },
    normal: { emoji: '🔵', label: 'Normal', color: '#0077b5' }
  };

  const { emoji, label, color } = config[type];

  badge.style.cssText = `
    position: absolute;
    top: 10px;
    right: 10px;
    background: ${color};
    color: white;
    padding: 8px 14px;
    border-radius: 20px;
    font-weight: 700;
  `;

  badge.innerHTML = `${emoji} ${label}`;
  card.appendChild(badge);
}

Performance Optimization

Initial version was slow (200ms per invitation). Optimizations:

Cache keyword processing: Lowercase keywords once, not on every check
Mark processed nodes: Use dataset.processed to avoid re-scanning
Debounce mutations: Don't process every single DOM change

Result: 15ms per invitation (13x faster)

Privacy by Design

No external API calls. No tracking. No data collection.

All settings stored locally using Chrome's Storage API:

chrome.storage.sync.set({ 
  settings: userSettings 
});

This syncs across user's devices via Chrome's encrypted sync.

Results

Week 1: 1,000+ users
Week 2: 3,000+ users
Average rating: 4.8/5

Top feedback: "This should be built into LinkedIn"

What I Learned

Simple > Complex: I almost built an ML classifier. Keyword matching works for 85% of cases.
Performance matters: Users notice 200ms delays. They don't notice 15ms.
Privacy sells: "No tracking" was the most-mentioned feature in reviews.
Distribution is everything: Building took 16 hours. Marketing took 40 hours.

Try It Yourself

Chrome Web Store: https://chromewebstore.google.com/detail/Request%20Radar/pmpkcbnkoojpenphcempidfgkjkemife

What's Next?

Planning to add:

Analytics dashboard (local-only)
Better classification (context-aware)
Premium tier ($3.99/mo) with advanced features

Build Your Own

Want to create a similar extension? Here's the tech stack:

Vanilla JavaScript (no frameworks)
Chrome Extension Manifest V3
MutationObserver API
Chrome Storage API

For me, it wasn't "detect spam" - it was "don't miss opportunities."

What obvious-but-unbuilt tool will you create?

Questions? Drop them in the comments!

DEV Community: Kevin Tran

I Built a LinkedIn Spam Filter in a Weekend (And You Can Too)

I Built a LinkedIn Spam Filter in a Weekend (And You Can Too)

The Problem

The Solution

How It Works

1. Content Script Injection

2. Detecting New Invitations

3. Classification Logic

4. Adding Visual Badges

Performance Optimization

Privacy by Design

Results

What I Learned

Try It Yourself

What's Next?

Build Your Own