Hypertweet

AI-powered browser extension that generates intelligent, contextual replies on Twitter/X, Reddit, and LinkedIn.

What It Does

Hypertweet reads the conversation you're replying to and generates contextually-aware responses using AI. Pick a tone, click once, and get a draft you can edit before posting.

Supported Platforms:

• X (Twitter)
• Reddit
• LinkedIn

Features

• 5 Built-in Tones - Professional, Friendly, Witty, Insightful, Casual
• Custom Tones - Create your own with free-form AI instructions
• Thread-Aware - Understands full conversation context, not just the parent post
• Platform-Specific - Follows character limits, formatting norms, and culture per site
• Model Selection - Choose from multiple models via Groq

Architecture

┌─────────────────────────────────────────────────────────┐
│                    Browser Extension                     │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────────┐  │
│  │   Twitter   │  │   Reddit    │  │    LinkedIn     │  │
│  │   Scraper   │  │   Scraper   │  │     Scraper     │  │
│  └──────┬──────┘  └──────┬──────┘  └────────┬────────┘  │
│         └────────────────┼──────────────────┘           │
│                    ┌─────┴─────┐                        │
│                    │  React UI │                        │
│                    └─────┬─────┘                        │
└──────────────────────────┼──────────────────────────────┘
                           │ HTTPS
                     ┌──────┴──────┐
                     │    Go API   │
                     │   Server    │
                     └──────┬──────┘
                            │
               ┌────────────┼────────────┐
               │            │            │
         ┌─────┴─────┐ ┌────┴────┐ ┌─────┴─────┐
         │  MongoDB  │ │   JWT   │ │   Groq AI │
         │           │ │  Auth   │ │           │
         └───────────┘ └─────────┘ └───────────┘

Tech Stack

Component	Technology
Extension	TypeScript, React 19, Emotion, React Query
Build	Bun, esbuild
Server	Go 1.22+ (stdlib net/http)
Database	MongoDB
Auth	JWT + BCrypt
AI	Groq API

Development

Prerequisites

• Bun
• Go (1.22+)
• MongoDB

Extension

cd extension

# Install dependencies
just install

# Development with hot reload
just dev

# Production build
just prod

# Run all checks (lint, typecheck, format)
just check

Load the unpacked extension from extension/dist in Chrome.

Server

cd server

# Required
export GROQ_API_KEY="your-groq-api-key"

# Run development server
make run

# Run tests
make test

Available Commands

Extension (justfile):

Command	Description
just install	Install dependencies
just watch	Build + watch with auto-reload
just build	Production build
just dev	Clean + watch
just prod	Clean + production build
just format	Prettier formatting
just lint	ESLint with auto-fix
just typecheck	TypeScript validation
just check	All validation checks
just clean	Remove dist directory

Project Structure

v2/
├── extension/              # Chrome extension
│   ├── src/
│   │   ├── content.ts      # Entry point, injected into pages
│   │   ├── router.ts       # Platform detection
│   │   ├── base.ts         # Abstract scraper interface
│   │   ├── twitter.ts      # X/Twitter scraper
│   │   ├── reddit.ts       # Reddit scraper
│   │   ├── linkedin.ts     # LinkedIn scraper
│   │   ├── models.ts       # Data types (Page, Post, User)
│   │   ├── api.ts          # Server API client
│   │   └── ui/             # React components
│   │       ├── Keyboard.tsx    # Main interaction UI
│   │       ├── AuthForm.tsx    # Login/register
│   │       ├── ToneEditor.tsx  # Custom tone creation
│   │       └── components/     # Shared UI primitives
│   ├── manifest.json
│   └── justfile
│
└── server/                 # Go backend
    ├── cmd/server/         # Server entrypoint
    └── internal/           # Feature packages
        ├── auth/           # Login, register, refresh (JWT)
        ├── profiles/       # Model selection, chat config
        ├── tones/          # Tone CRUD + defaults
        ├── apitokens/      # MCP token CRUD (revocable)
        ├── mcp/            # MCP endpoint (/mcp)
        └── ai/             # Groq integration

How It Works

1. Scrape - Extension parses the current page, extracting the post you're replying to plus full thread context

2. Generate - Click a tone button to send the context + selected tone to the server

3. Prompt - Server builds a platform-aware prompt with:

   • Site-specific guidelines (character limits, culture, formatting)
   • Tone instruction (e.g., "Be professional and concise")
   • Thread conversation (nested replies, truncated for token limits)
   • Active post content

4. Insert - AI-generated reply is inserted into the reply textarea for editing before posting

Configuration

Server configuration via environment variables:

Variable	Description
MONGODB_URI	MongoDB connection string
JWT_SECRET	Secret key for JWT signing
GROQ_API_KEY	Groq API key

Claude Code MCP

Hypertweet exposes a Model Context Protocol (MCP) server at POST /mcp for managing tones and profile/chat settings.

1. In the extension UI, open Settings → MCP and generate a token.
2. Add the MCP server to Claude Code.

Example .mcp.json entry (project scope):

{
  "mcpServers": {
    "hypertweet": {
      "type": "http",
      "url": "http://localhost:5001/mcp",
      "headers": {
        "Authorization": "Bearer ${HYPERTWEET_MCP_TOKEN}"
      }
    }
  }
}

Then run Claude Code with the env var set:

• HYPERTWEET_MCP_TOKEN=<paste-token-here> claude

Available AI Models

Configured in server/internal/profiles/types.go.

• openai/gpt-oss-120b (default)
• openai/gpt-oss-20b
• llama-3.3-70b-versatile
• llama-3.1-8b-instant

License

MIT