ReplyQueue

Stop scrolling past engagement opportunities. ReplyQueue watches your LinkedIn feed and surfaces posts that align with your blog content, so you can join conversations that matter.

Install from Chrome Web Store →

Why ReplyQueue?

Content creators spend hours scrolling social feeds hoping to find relevant discussions. ReplyQueue flips that model: connect your RSS feed, browse normally, and let AI surface the posts worth your time.

• Passive Discovery - Extracts posts as you scroll, no extra workflow
• Smart Matching - Keyword and AI-powered relevance scoring against your content
• Reply Suggestions - Generates responses that match your writing style
• View Post - One click to open matched posts in a new tab
• Dark Mode - System, light, or dark theme with one-click toggle
• Privacy-First - No analytics, no tracking, your data stays local
• Extensible - Platform adapter pattern makes adding new networks straightforward

Installation

From Chrome Web Store (Recommended)

Install ReplyQueue →

After installation:

1. Click the puzzle piece icon in Chrome toolbar
2. Pin ReplyQueue for easy access
3. Click the ReplyQueue icon to open the side panel and start setup

From Source (Development)

1. Clone the repository

   git clone https://github.com/charlesjones-dev/replyqueue.git
   cd replyqueue

2. Install dependencies

   pnpm install

3. Build the extension

   pnpm build

4. Load in Chrome

   • Open Chrome and navigate to chrome://extensions/
   • Enable "Developer mode" (toggle in top right)
   • Click "Load unpacked"
   • Select the dist folder from the project directory

5. Pin the extension

   • Click the puzzle piece icon in Chrome toolbar
   • Pin ReplyQueue for easy access

---

Development Setup

Prerequisites

• Node.js 18+
• pnpm 9+
• Chrome browser

Commands

# Start development server with hot reload
pnpm dev

# Run all tests
pnpm test

# Run tests in watch mode
pnpm test:watch

# Build for production
pnpm build

Loading the Development Build

1. Run pnpm dev to start the development server
2. Open chrome://extensions/
3. Enable Developer mode
4. Click "Load unpacked" and select the dist folder
5. The extension will hot-reload as you make changes

---

Architecture

ReplyQueue uses a modular architecture designed for cross-platform extensibility.

src/
├── platforms/           # Platform adapters (LinkedIn, future: Twitter, etc.)
│   ├── types.ts         # PlatformAdapter interface
│   ├── index.ts         # Platform registry
│   ├── linkedin/        # LinkedIn-specific implementation
│   └── _template/       # Template for adding new platforms
├── content/             # Content script (runs on social media pages)
│   ├── index.ts         # Entry point
│   ├── dom-observer.ts  # MutationObserver for feed changes
│   └── post-extractor.ts# Extracts posts using platform adapters
├── background/          # Service worker
│   ├── index.ts         # Message handling and orchestration
│   ├── rss-fetcher.ts   # RSS feed parsing and caching
│   ├── matcher.ts       # Keyword-based matching
│   └── openrouter.ts    # AI API client for semantic matching
├── sidepanel/           # Side panel UI (Vue 3)
│   ├── App.vue          # Root component
│   ├── views/           # SetupView, MainView, SettingsView
│   ├── components/      # Reusable UI components
│   └── composables/     # Vue composables for state management
└── shared/              # Shared utilities
    ├── types.ts         # TypeScript interfaces
    ├── constants.ts     # Configuration constants
    ├── storage.ts       # Chrome storage wrappers
    ├── validation.ts    # Input validation
    └── messages.ts      # Message passing types

Data Flow

1. Content Script extracts posts from LinkedIn feed using the platform adapter
2. Background Worker receives posts via message passing and stores them
3. RSS Fetcher periodically fetches and parses your blog's RSS feed
4. Matcher compares extracted posts against RSS content using keywords and AI
5. Side Panel displays matched posts with relevance scores and reply suggestions

---

Adding a New Platform Adapter

ReplyQueue is designed to support multiple social media platforms. To add a new platform:

1. Copy the template directory

   cp -r src/platforms/_template src/platforms/twitter

2. Update selectors.ts

   • Define CSS selectors for the platform's DOM structure
   • Include selectors for posts, authors, content, engagement metrics

3. Implement the adapter in adapter.ts

   • Update class name, platformId, and platformName
   • Implement extractPost() to parse the platform's DOM
   • Implement getPostUrl() for generating permalinks
   • Implement scrollToPost() for navigation

4. Register the platform in src/platforms/index.ts

   import { TwitterAdapter } from './twitter/adapter'
   
   const adapters: PlatformAdapter[] = [
     new LinkedInAdapter(),
     new TwitterAdapter(),
   ]

5. Update manifest.json

   • Add the platform's URL patterns to host_permissions
   • Add content script matches

6. Add tests

   • Create tests/platforms/twitter-extraction.test.ts
   • Test post extraction with mock DOM elements

7. Update allowed origins for message validation

   • In src/background/index.ts, add the platform's domain to ALLOWED_CONTENT_SCRIPT_ORIGINS

See CONTRIBUTING.md for detailed guidelines.

---

Configuration Options

Matching Preferences

Option	Default	Options	Description
Relevance Threshold	30%	10-90%	Minimum score for a post to be considered a match
Max Posts to Show	20	10, 20, 30, 50, 100	Maximum number of matched posts to display
RSS Cache Duration	60 min	15, 30, 60, 120, 240	How long to cache RSS feed before refreshing
Max RSS Items	25	10, 25, 50, 100	Maximum items to fetch from RSS feed
Max Blog Items in AI Prompts	25	10, 25, 50, All	Blog items included when AI evaluates posts
Blog Content Sent to AI	2,500 chars	1K, 2.5K, 5K, 10K, No limit	Amount of blog content included when matching
Social Post Content Sent to AI	1,000 chars	500, 1K, 2K, 3K, No limit	Amount of post content included when evaluating

AI Settings

Option	Default	Description
Model	anthropic/claude-haiku-4.5	OpenRouter model for semantic matching and reply generation
Writing Style Examples	-	Up to 10 examples of your previous comments to match your tone
Communication Preferences	-	Custom writing rules for AI-generated replies (e.g., formatting, tone)

---

Privacy & Data Handling

ReplyQueue is built with privacy as a core principle.

What stays on your device:

• All extracted post data
• Your configuration and preferences
• Writing style examples
• Cached RSS content

What is sent externally:

• OpenRouter API - Post content and RSS data for AI matching (only when you trigger a scan)
• Your RSS Feed URL - Fetched periodically to get your blog content

No tracking:

• No analytics
• No telemetry
• No usage data collection
• No third-party scripts

See privacy-policy.md for the full privacy policy.

---

Security

ReplyQueue uses multiple layers of security tooling to catch vulnerabilities early.

Static Analysis

Tool	Purpose	Command
eslint-plugin-security	Detects common security anti-patterns in JS/TS	pnpm lint
pnpm audit	Scans dependencies for known CVEs	pnpm audit:check
Semgrep	SAST scanning for injection, XSS, and more	CI only (or Docker locally)

CI Integration

The .github/workflows/security.yml workflow runs Semgrep on every push and PR to main, using:

• auto - Semgrep's recommended security rules
• p/javascript - JavaScript-specific security patterns
• p/typescript - TypeScript-specific security patterns

Running Security Checks Locally

# ESLint security rules (included in lint)
pnpm lint

# Dependency vulnerability scan
pnpm audit:check

# Semgrep via Docker
docker run --rm -v "$(pwd):/src" semgrep/semgrep semgrep scan --config auto --config p/javascript --config p/typescript /src

Security Practices

• Origin validation - Background script validates message origins against allowlist
• Input sanitization - All user inputs validated before processing
• No eval/Function - Dynamic code execution patterns are blocked by ESLint
• Secure storage - API keys stored in chrome.storage.local, never exposed to content scripts
• CSP compliance - Extension follows Chrome's Content Security Policy requirements

---

Troubleshooting

Extension not loading

• Ensure you're loading the dist folder, not the project root
• Check for build errors: pnpm build
• Look for errors in chrome://extensions/

Posts not being extracted

• Make sure you're on linkedin.com
• Scroll through the feed to trigger extraction
• Check the console for errors (right-click extension icon > "Inspect")

API key validation failing

• Verify your OpenRouter API key starts with sk-or-v1-
• Check that your API key has credits available
• Try the key at openrouter.ai

RSS feed not loading

• Ensure the URL returns valid RSS/Atom XML
• Try the URL in a browser to verify it's accessible
• Check for CORS issues in the console

Matches not appearing

• Lower the relevance threshold in settings
• Ensure your RSS feed has recent content
• Try refreshing the feed manually

---

Roadmap

Planned Platforms

• Twitter/X
• Mastodon
• Bluesky
• Reddit

Future Features

• Batch reply generation
• Reply history tracking
• Multiple RSS feed support
• Custom matching prompts
• Export matches to CSV
• Browser notifications for high-relevance matches

---

Contributing

Contributions are welcome! Please read CONTRIBUTING.md for guidelines on:

• Setting up the development environment
• Code style and linting
• Submitting issues and pull requests
• Adding new platform adapters

---

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Acknowledgments

• OpenRouter for AI model access
• CRXJS for Vite Chrome extension integration
• Vue.js for the reactive UI framework