PromptForge AI 🚀

Professional Coding Prompt Generator & Technical Spec Compiler (Chrome Extension)

PromptForge AI is a privacy-first, developer-focused Google Chrome extension (Manifest V3) that turns casual, vague coding ideas into production-grade, highly structured technical prompts and architecture specifications for AI coding tools like ChatGPT, Claude, Gemini, Cursor, Copilot, and GitHub.

Now featuring an instant, zero-key multi-model failover engine: PromptForge AI works right after installation with no API keys required, powered by LLaMA 3.3 70B with transparent self-healing failover to Gemini 3.8 Flash, Gemini 2.5 Flash, and an Offline Deterministic Synthesizer.

🌟 Why PromptForge AI?

Vague prompts lead to hallucinated logic, missing edge cases, and incomplete boilerplate when prompting modern LLMs. PromptForge AI bridges this gap by acting as an intelligent prompt compilation and refinement layer:

Zero-Key Out of the Box: Start generating high-precision coding prompts immediately upon installation—no setup gates or required API keys.
Multi-Model Self-Healing Failover: Cascades across high-speed cloud LLMs and an offline synthesizer so your workflow is never interrupted by rate limits or network dropouts.
Structure & Precision: Enforces architectural context, technical stack constraints, edge cases, error handling, security considerations, and concrete acceptance criteria.
Side Panel Workflow: Lives right alongside your code editors and browser tabs using Chrome's native Side Panel API (chrome.sidePanel).
In-Page Magic Wand: Seamlessly detects prompt inputs on ChatGPT, Claude, Gemini, and GitHub, enhancing drafts in-place with a single click.
100% Privacy & Optional BYOK: Zero third-party proxy servers. Your prompts and optional custom API keys are stored locally in your browser (chrome.storage.local) and communicate directly with official provider endpoints.

⚡ Zero-Key Multi-Model Failover Architecture

PromptForge AI features an autonomous, multi-tier cascade engine designed to maximize speed, quality, and reliability:

┌────────────────────────────────────────────────────────┐
│               User Prompt Enhancement Request          │
└───────────────────────────┬────────────────────────────┘
                            │
                            ▼
      ┌───────────────────────────────────────────┐
      │  ⚡ Tier 1: LLaMA 3.3 70B (Groq Cloud)    │ ───► Success (Ultra-fast ~300ms)
      │  `llama-3.3-70b-versatile`                │
      └─────────────────────┬─────────────────────┘
                            │ (429 Rate Limit / 5xx Error)
                            ▼
      ┌───────────────────────────────────────────┐
      │  ✨ Tier 2: Gemini 3.8 Flash              │ ───► Success (Google AI Studio)
      │  `gemini-3.8-flash`                       │
      └─────────────────────┬─────────────────────┘
                            │ (429 Quota / 5xx Error)
                            ▼
      ┌───────────────────────────────────────────┐
      │  ✨ Tier 3: Gemini 2.5 Flash              │ ───► Success (High Quota Cloud)
      │  `gemini-2.5-flash`                       │
      └─────────────────────┬─────────────────────┘
                            │ (Network Offline / Service Error)
                            ▼
      ┌───────────────────────────────────────────┐
      │  🛡️ Tier 4: Offline Deterministic Engine  │ ───► Instant On-Device Spec
      │  Zero-network structural prompt compiler  │
      └───────────────────────────────────────────┘

⚡ GPT-OSS 120B & LLaMA 3.3 70B (openai/gpt-oss-120b / llama-3.3-70b-versatile via Groq Cloud):
- Primary Engine: Ultra-fast inference (~300 tokens/sec), 131,072-token context window, deep architectural reasoning, and automatic model tier failover across 120B, 70B, and 27B models.
✨ Gemini 3.8 Flash (gemini-3.8-flash via Google Generative Language):
- First Automatic Fallback: Advanced multimodal reasoning model that seamlessly picks up requests if Groq encounters rate limits (HTTP 429) or transient 5xx server errors in under 50ms.
✨ Gemini 2.5 Flash (gemini-2.5-flash):
- Second Automatic Fallback: Ultra-high quota model ensuring cloud generation redundancy under heavy traffic.
🛡️ Domain-Intelligent Offline Synthesizer:
- Zero-Network Safeguard: Generates comprehensive, production-grade 3-paragraph master prompts fully on-device without quoting the user's raw input. Features specialized domain architectural intelligence for telemetry/weather, auth/security, web scraping, and full-stack systems.
🌐 Optional Cloud Relay Proxy (Cloudflare Worker):
- Includes a production-ready edge worker (server/) that can be deployed to Cloudflare Workers with npx wrangler deploy to provide a private, zero-key relay endpoint for your team or organization.
🏷️ Real-Time Model Badges:
- The Studio output viewer displays dynamic badges (e.g. ⚡ GPT-OSS 120B, ⚡ LLaMA 3.3, ✨ Gemini 3.8, ✨ Gemini 2.5, 🌐 LLaMA Proxy, or 🛡️ Local Synthesizer) so you always know which engine fulfilled your request.

✨ Features

1. 🎛️ Side Panel Studio (`chrome.sidePanel`)

Interactive Prompt Studio: Input casual ideas, specify target tech stacks (e.g. Next.js, Node.js, FastAPI, Rust, Go, Python), select LLM targets, and adjust detail level.
Template Library: Curated, battle-tested prompt templates:
- Full-Stack Feature Specification
- System Architecture & API Design
- Unit & Integration Test Suite
- Database Schema & Migration
- Bug Investigation & Root Cause Analysis
- Refactoring & Code Modernization
Local Prompt History: Searchable, timestamped history saved in chrome.storage.local with one-click copy, active model metadata, and instant reload into the editor.
Copy & Export: Instant Markdown copy to clipboard with visual toast confirmations.

2. 🪄 In-Page Magic Wand

Automatically attaches an interactive magic wand button to text inputs on:
- ChatGPT (https://chatgpt.com/*)
- Claude (https://claude.ai/*)
- Google Gemini (https://gemini.google.com/*)
- GitHub Discussions / Issues / PRs (https://github.com/*)
Click the wand to compile your rough thought into an exhaustive technical prompt without ever leaving the page.

3. 🧠 Smart Auto-Detection: Kickoff vs. Follow-Up Mini-Prompts

Never get stuck repeating "You are a senior software engineer acting as technical lead..." when you're in the middle of a coding conversation:

🎯 Project Kickoff: When architecting a new feature or service ("make a python weather app"), PromptForge AI compiles a comprehensive, multi-paragraph Master Architectural Specification.
⚡ Follow-Up Steer: When providing iterative steering ("make it faster", "now do step 2", "TypeError on line 14", "add pytest tests"), PromptForge AI automatically detects conversational follow-up intent and outputs a surgical, high-leverage mini-prompt with ZERO persona intro boilerplate.
🏷️ Real-Time Visual Feedback: Live intent indicators (🎯 Project Kickoff vs ⚡ Follow-Up Steer) in both the Chrome Extension Side Panel and the Web Studio.

4. 🔑 Bring Your Own Key (BYOK) — Optional Power Mode

While PromptForge AI works out of the box with zero keys, power users can configure custom API keys in the Settings tab for direct access:
- Groq Cloud (LLaMA 3.3 70B, LLaMA 3.1 8B)
- Google AI Studio (Gemini 3.8 Flash, Gemini 2.5 Flash, Gemini 2.5 Pro)
- OpenAI (GPT-4o, GPT-4o-mini)
- Anthropic (Claude 3.5 Sonnet, Claude 3.5 Haiku)
Direct HTTPS calls from the extension background service worker with zero middleman or telemetry.

🚀 Quick Start Guide (Load Unpacked in Chrome)

You can load and use PromptForge AI immediately in Google Chrome, Brave, Arc, or Microsoft Edge:

Prerequisites

Node.js (version 18 or later recommended)
npm (bundled with Node.js)

Step 1: Clone and Install Dependencies

git clone https://github.com/JISHU-GHOSH/prompt-generator.git
cd prompt-generator
npm install

Step 2: Build the Extension

npm run build

This runs TypeScript validation (tsc --noEmit) and compiles all extension bundles via Vite into the dist/ directory.

Step 3: Load into Chrome

Open Google Chrome and navigate to:
```
chrome://extensions
```
In the top-right corner, toggle Developer mode to ON.
In the top-left corner, click the Load unpacked button.
In the file picker dialog, select the dist directory (e.g. c:\Users\...\prompt-generator\dist).
Click Select Folder.
PromptForge AI is now loaded and ready! You will see PromptForge AI - Professional Coding Prompt Generator in your extensions list.

Step 4: Pin and Open PromptForge AI

Click the puzzle icon (Extensions menu) in the Chrome toolbar.
Find PromptForge AI and click the Pin icon.
Click the PromptForge AI icon to open the Side Panel Studio.
Type any rough idea (e.g. "JWT auth with refresh tokens in FastAPI") and click ⚡ Enhance with AI.
No API keys or configuration needed! The multi-model engine will synthesize a complete technical specification in milliseconds.

🔑 Optional: Configuring Personal API Keys

If you have dedicated developer keys with higher rate limits, you can plug them into the Settings tab:

1. Groq Cloud (Free high-speed LLaMA 3.3 70B)

Go to Groq Console.
Create and copy an API key (gsk_...).
In PromptForge AI Settings, expand Custom API Keys, select Groq, and paste your key.

2. Google AI Studio (Gemini 3.8 Flash / 2.5 Flash)

Go to Google AI Studio.
Sign in with your Google account and click Get API key.
Copy your Gemini API key.
In PromptForge AI Settings, paste your key under Google Gemini.

3. OpenAI (GPT-4o)

Go to OpenAI Platform API Keys.
Generate a secret key (sk-...).
In PromptForge AI Settings, select OpenAI, paste your API key, and click Save Settings.

4. Anthropic (Claude 3.5 Sonnet)

Go to Anthropic Console.
Generate a new key (sk-ant-...).
In PromptForge AI Settings, select Anthropic, paste your API key, and click Save Settings.

🛠️ Development & Testing

Available Commands

Command	Description
`npm run dev`	Starts Vite extension development server
`npm run dev:web`	Starts standalone Web Studio development server (`localhost:5173`)
`npm run build`	Compiles and packages Chrome Extension into `dist/`
`npm run build:web`	Compiles production Web Studio into `dist-web/`
`npm test`	Runs the full Vitest automated test suite (148+ tests)
`npm run generate-icons`	Generates official PNG icon set in `public/icons/`

🌐 Deploy as a Live Web Application (Vercel)

PromptForge AI includes a full-screen, responsive Dual-Pane Web Studio alongside the Chrome Extension:

Push this repository to your GitHub account (git push origin master).
Go to vercel.com and click "Add New Project" → Import prompt-generator.
Under Environment Variables, add:
- Key: GROQ_API_KEY
- Value: Your Groq API key (gsk_...)
Click Deploy!
- Your website is instantly live at https://your-project.vercel.app with zero-key serverless generation powered by openai/gpt-oss-120b!

Running the Test Suite

npm test

Runs 138+ automated unit and component tests verifying:

Failover Cascade: Transparent switching from LLaMA 3.3 70B to Gemini 3.8, Gemini 2.5, and local synthesizer on rate limits or errors
Groq & Gemini Clients: Streaming SSE parsing, model resolution, and status code classification
Manifest V3 Configuration: Manifest validation and bundled asset integrity
Background Service Worker: Message router, model tracking, and history persistence
Side Panel UI: Zero-key studio workflow, model execution badges, templates, and history
In-Page Content Script: Wand injection, site adapters (ChatGPT, Claude, Gemini, GitHub), and input synchronization
Storage Service: Chrome storage wrapper, settings persistence, and history management

📁 Project Structure

prompt-generator/
├── manifest.json              # Extension Manifest V3 configuration
├── vite.config.ts             # Vite multi-page build configuration
├── tailwind.config.js         # Tailwind CSS styling configuration
├── public/
│   └── icons/                 # Extension PNG icons (16px, 48px, 128px)
├── src/
│   ├── background/
│   │   └── service-worker.ts  # Background worker, failover router & sidepanel opener
│   ├── content/
│   │   ├── content-script.ts  # In-page magic wand overlay & site adapters
│   │   └── content.css        # Animations & styles for floating wand
│   ├── sidepanel/
│   │   ├── index.html         # HTML entry point for Chrome Side Panel
│   │   ├── main.tsx           # React root mount
│   │   ├── App.tsx            # Main application UI with tabs & state
│   │   ├── components/        # Studio, Templates, History, Settings, Badges
│   │   └── index.css          # Tailwind CSS styles
│   └── services/
│       ├── ai/                # FailoverRouter, Groq, Gemini, OpenAI & Anthropic clients
│       ├── prompt-engine/     # Meta-prompt compiler & template engine
│       └── storage.ts         # Chrome storage wrapper (chrome.storage.local)
├── tests/                     # 138+ automated Vitest tests
│   ├── ai-client.test.ts
│   ├── content-script.test.ts
│   ├── failover-router.test.ts
│   ├── groq-client.test.ts
│   ├── manifest.test.ts
│   ├── prompt-compiler.test.ts
│   ├── service-worker.test.ts
│   ├── sidepanel.test.tsx
│   └── storage.test.ts
└── dist/                      # Production-ready loadable extension package
    ├── manifest.json
    ├── src/
    │   ├── background/service-worker.js
    │   ├── content/content-script.js
    │   ├── content/content.css
    │   └── sidepanel/index.html
    ├── assets/
    └── icons/

🔒 Security & Privacy

No Third-Party Backend: PromptForge AI operates directly from the browser to provider APIs with zero middleman proxy.
Secure Key Storage: All settings and optional API keys are stored in chrome.storage.local within your browser's encrypted profile directory.
Zero-Key Privacy: In default zero-key mode, prompt compilation requests flow directly to fast provider endpoints without storing any personal identifiers.
Scoped Permissions: Requests only sidePanel, storage, and activeTab permissions, plus explicitly defined host permissions.

📄 License

Distributed under the MIT License. See LICENSE for more information.

Name		Name	Last commit message	Last commit date
api		api
docs/superpowers		docs/superpowers
public/icons		public/icons
scripts		scripts
server		server
src		src
tests		tests
.gitignore		.gitignore
CHROMEWEBSTORE.md		CHROMEWEBSTORE.md
PRIVACY.md		PRIVACY.md
README.md		README.md
index.html		index.html
manifest.json		manifest.json
package-lock.json		package-lock.json
package.json		package.json
postcss.config.js		postcss.config.js
tailwind.config.js		tailwind.config.js
tsconfig.json		tsconfig.json
vercel.json		vercel.json
vite.config.ts		vite.config.ts
vite.config.web.ts		vite.config.web.ts

Navigation Menu

Repository navigation

About

Repository files navigation

PromptForge AI 🚀

Professional Coding Prompt Generator & Technical Spec Compiler (Chrome Extension)

🌟 Why PromptForge AI?

⚡ Zero-Key Multi-Model Failover Architecture

✨ Features

1. 🎛️ Side Panel Studio (`chrome.sidePanel`)

2. 🪄 In-Page Magic Wand

3. 🧠 Smart Auto-Detection: Kickoff vs. Follow-Up Mini-Prompts

4. 🔑 Bring Your Own Key (BYOK) — Optional Power Mode

🚀 Quick Start Guide (Load Unpacked in Chrome)

Prerequisites

Step 1: Clone and Install Dependencies

Step 2: Build the Extension

Step 3: Load into Chrome

Step 4: Pin and Open PromptForge AI

🔑 Optional: Configuring Personal API Keys

1. Groq Cloud (Free high-speed LLaMA 3.3 70B)

2. Google AI Studio (Gemini 3.8 Flash / 2.5 Flash)

3. OpenAI (GPT-4o)

4. Anthropic (Claude 3.5 Sonnet)

🛠️ Development & Testing

Available Commands

🌐 Deploy as a Live Web Application (Vercel)

Running the Test Suite

📁 Project Structure

🔒 Security & Privacy

📄 License

About

Releases

Packages

Contributors

Languages

Latest commit

History

Folders and files

Repository files navigation

PromptForge AI 🚀

Professional Coding Prompt Generator & Technical Spec Compiler (Chrome Extension)

🌟 Why PromptForge AI?

⚡ Zero-Key Multi-Model Failover Architecture

✨ Features

1. 🎛️ Side Panel Studio (chrome.sidePanel)

2. 🪄 In-Page Magic Wand

3. 🧠 Smart Auto-Detection: Kickoff vs. Follow-Up Mini-Prompts

4. 🔑 Bring Your Own Key (BYOK) — Optional Power Mode

🚀 Quick Start Guide (Load Unpacked in Chrome)

Prerequisites

Step 1: Clone and Install Dependencies

Step 2: Build the Extension

Step 3: Load into Chrome

Step 4: Pin and Open PromptForge AI

🔑 Optional: Configuring Personal API Keys

1. Groq Cloud (Free high-speed LLaMA 3.3 70B)

2. Google AI Studio (Gemini 3.8 Flash / 2.5 Flash)

3. OpenAI (GPT-4o)

4. Anthropic (Claude 3.5 Sonnet)

🛠️ Development & Testing

Available Commands

🌐 Deploy as a Live Web Application (Vercel)

Running the Test Suite

📁 Project Structure

🔒 Security & Privacy

📄 License

About

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages

1. 🎛️ Side Panel Studio (`chrome.sidePanel`)