LinkedIn Reply Assistant

Draft polished replies to recruiter messages in one click — no API, no account, nothing leaves your browser.

The Problem

If you're active on LinkedIn, you know the drill: dozens of recruiter messages every week asking if you're open to new opportunities or if you know someone who is. Replying to each one manually is tedious. Ignoring them feels rude.

LinkedIn Reply Assistant fixes this. It sits quietly on your LinkedIn messaging page, reads incoming messages, figures out what the recruiter wants, and drafts a polished reply in the right language. You review it, click insert, done.

No AI service. No API key. No data sent anywhere. Everything runs locally in your browser.

Features

• Smart intent detection — classifies messages as job offer, referral request, or other using 80+ weighted keywords in French and English
• Automatic language detection — detects whether the recruiter wrote in French or English and replies in the same language; force either with one click
• One-click replies — copy to clipboard, insert directly into the LinkedIn message box, or unlock the text area to edit before sending
• Fully customizable templates — edit every template from the popup; use {firstName}, {name}, and {userName} variables for personalization
• Confidence score — shows how confident the classifier is in its intent detection
• Privacy first — zero API calls, zero tracking, zero accounts; runs entirely in your browser

How It Works

Open a LinkedIn conversation
         │
         ▼
Extension reads the latest messages
         │
         ▼
Classifier scores keywords (80+ terms, FR + EN)
  ├── Job offer?         → drafts a decline reply
  ├── Referral request?  → drafts a "no one in mind" reply
  └── Other?             → drafts a generic reply
         │
         ▼
Language detected (French or English)
         │
         ▼
Reply panel appears — you review and click Insert

All of this runs locally. No network requests are made.

Installation

Works in Chrome, Edge, Brave, and Arc (any Chromium-based browser).

Step 1: Get the code

git clone https://github.com/End2EndAI/linkedin-auto-reply.git

Or click the green Code button above → Download ZIP → unzip anywhere on your computer.

Step 2: Load into your browser

1. Open chrome://extensions/ (or edge://extensions/, brave://extensions/)
2. Toggle Developer mode ON (top-right corner)
3. Click Load unpacked
4. Select the linkedin-auto-reply folder
5. The extension icon appears in your toolbar

Step 3: Set your name

Click the extension icon in the toolbar, go to the Settings tab, and enter your name. This populates the {userName} variable in reply templates.

Step 4: Go to LinkedIn

Open linkedin.com/messaging — a small robot button will appear in the bottom-right corner. The panel opens automatically when a recruiter message is detected.

Usage

Action	What happens
Open a conversation	Extension automatically analyzes the latest messages
Click the robot button	Opens the assistant panel manually
Click Copy	Copies the suggested reply to your clipboard
Click Insert	Pastes the reply directly into the LinkedIn message box
Click Edit	Unlocks the text area so you can tweak the reply
Click a flag (🇫🇷 / 🇬🇧)	Forces the response language
Click an intent button (Job / Referral / Other)	Overrides the detected intent

Customizing Your Replies

Click the extension icon in your Chrome toolbar to open the popup:

Settings tab — Set your name (used in {userName}) and toggle auto-detection on or off.

French / English tabs — Edit each template separately. Three templates per language: job offer, referral request, and other.

Available template variables:

Variable	Replaced with
{firstName}	Recruiter's first name
{name}	Recruiter's full name
{userName}	Your name (set in Settings)

Stats tab — See how many messages the extension has analyzed and how many replies you've copied or inserted.

FAQ

Will this get my LinkedIn account banned? No. The extension only reads the page content and helps you compose replies. It does not automate sending messages, scrape data, or interact with LinkedIn's API. You always press send manually.

Does it use AI / ChatGPT / any API? No. Classification is done locally with a keyword scoring algorithm. No external service, no API key, no cost, no data sent anywhere.

What languages are supported? French and English. The extension auto-detects which one to use based on the incoming message. You can also force either language with the flag buttons.

Does it work with LinkedIn Premium / Sales Navigator? Yes. It works on any page under linkedin.com/messaging.

The panel isn't appearing — what's wrong? A few things to check:

• Make sure you're on linkedin.com/messaging (not linkedin.com/in/...)
• The extension polls every 2 seconds — wait a moment after opening a conversation
• If the message scores below the confidence threshold, the panel won't auto-appear; click the robot button to open it manually
• If LinkedIn updates its DOM structure, selectors may need updating — open an issue

Can I add more languages? Yes — see CONTRIBUTING.md for a step-by-step guide on adding a new language. PRs welcome.

Tech Stack

Component	Technology
Extension format	Chrome Manifest V3
Language	Vanilla JavaScript (zero dependencies)
Classifier	Weighted keyword scoring (80+ terms, FR + EN)
Language detection	Word-frequency analysis
DOM observation	MutationObserver + interval polling
Storage	Chrome Storage Sync API
Styling	Custom CSS with dark mode support

Project Structure

linkedin-auto-reply/
├── manifest.json       Chrome extension config
├── classifier.js       Intent + language classifier
├── content.js          Main script injected on LinkedIn
├── content.css         Floating panel styles + dark mode
├── background.js       Service worker, storage defaults
├── popup.html          Extension popup UI
├── popup.js            Popup logic (settings, templates, stats)
├── popup.css           Popup styles
├── assets/             Screenshots and logo for documentation
└── icons/              Extension icons (16, 48, 128px)

Contributing

Contributions are welcome — see CONTRIBUTING.md for how to get started.

Some ideas if you're looking for something to work on:

• Add more languages (German, Spanish, Portuguese...)
• Improve the classifier with n-gram analysis
• Add support for LinkedIn InMail detection
• Create a "snooze" feature to remind you to reply later
• Add a keyboard shortcut to trigger the panel

License

MIT License. See LICENSE for details.

---

Built by End2EndAI with the help of ❤️ Claude Code ❤️