⚡ PromptGod

Supercharge, Enhance, and Deploy AI Prompts in Real-Time Across Any Generative AI Platform.

PromptGod is a next-generation, agentic Chrome Extension (Manifest V3) engineered to elevate prompt crafting directly inside any web app. By combining multi-agent LLM reasoning, real-time Server-Sent Events (SSE) streaming, dual-layer cloud/local caching, and robust React/Lexical DOM insertion, PromptGod transforms rough, casual thoughts into precision-engineered, production-grade AI prompts without ever leaving your workflow.

---

🌟 Key Features

• Autonomous 3-Tier Multi-Agent System:
  • Agent 1 (Site Mapper): Scans the active web page and discovers all available AI generative capabilities (e.g. Text-to-Image, Code Generation, Copywriting, Video Prompting).
  • Agent 2 (Meta-Prompt Architect): Generates specialized system prompts and keyword rules tailored specifically to the target platform and generative function.
  • Agent 3 (Real-Time Enhancer): Streams detailed, high-fidelity prompt expansions directly to your screen using Server-Sent Events (SSE).
• Dual-Layer Cloud & Local Caching:
  • Cloud (Google Firestore REST API): Global crowd-sourced cache for site maps and function prompts. If a site or feature has already been mapped by any user, Agents 1 and 2 are skipped, resulting in near-instant prompt generation.
  • Local (Chrome Storage): Instant zero-latency local fallback cache.
• Multi-Provider & Cutting-Edge Model Support:
  • Google AI Studio: gemma-4-31b-it, gemma-4-26b-a4b-it, gemini-2.5-flash-lite, gemini-3.5-flash-lite, gemini-3.1-flash-lite, + Custom IDs.
  • OpenRouter (Free & Paid Tier): nex-agi/nex-n2.5-mini:free, liquid/lfm-2.5-2.6b:free, nvidia/nemotron-3-embed-1b:free, nvidia/nemotron-3-ultra-550b-a55b:free, google/gemma-4-26b-a4b-it:free, google/gemma-4-31b-it:free, nvidia/nemotron-3-super-120b-a12b:free, + Custom IDs.
  • Groq (Ultra-Fast Inference): openai/gpt-oss-20b, openai/gpt-oss-120b, llama-3.3-70b-versatile, llama-3.1-8b-instant, groq/compound-mini, groq/compound, qwen/qwen3.8-27b, qwen/qwen3.6-27b, + Custom IDs.
• True Glassmorphic Shadow DOM UI:
  • Injected inside an isolated Shadow Root—100% immune to CSS pollution from host websites.
  • Live Draft Area: Inspect and fine-tune your captured prompt before triggering the enhancement.
  • Streaming Preview: Watch the AI stream words token-by-token.
  • One-Click Actions: Instant clipboard Copy, Regenerate, and Confirm & Paste.
• Bulletproof React & Lexical DOM Insertion:
  • Compatible with complex modern editors (WhatsApp Web, ChatGPT, Claude, Midjourney, Discord, Slack, etc.).
  • Uses native InputEvent dispatching and document.execCommand('insertText') to preserve React internal state, undo trees, and history without destructive DOM mutation.
• Multi-Frame & Iframe Intelligence:
  • Dynamically detects and tracks focused input frames across all_frames and iframe boundaries.

---

🏗️ Architecture Overview

flowchart TD
    User["User Selection / Hotkey (Ctrl+Shift+P)"] --> ExtTrigger["PromptGod Trigger (content.js)"]
    ExtTrigger --> FrameDetect["Frame & Input Capture (Native Shadow DOM UI)"]
    
    subgraph AgenticPipeline ["Multi-Agent Pipeline (background.js)"]
        FrameDetect --> CacheCheck1{"Firestore / Local Cache: Site Map?"}
        CacheCheck1 -- "Hit" --> FuncList["Display Function Grid"]
        CacheCheck1 -- "Miss" --> Agent1["Agent 1: Site Feature Discovery"]
        Agent1 --> SaveMap["Save Map to Firestore & Local"]
        SaveMap --> FuncList
        
        FuncList --> UserSelect["User Selects Function & Edits Draft"]
        UserSelect --> CacheCheck2{"Firestore / Local Cache: Function Prompt?"}
        CacheCheck2 -- "Hit" --> Agent3["Agent 3: SSE Streaming Enhancer"]
        CacheCheck2 -- "Miss" --> Agent2["Agent 2: Meta-Prompt Architect"]
        Agent2 --> SaveSysPrompt["Save Rule to Firestore & Local"]
        SaveSysPrompt --> Agent3
    end

    subgraph OutputFlow ["Stream & Action"]
        Agent3 --> StreamUI["Token Stream to UI Preview"]
        StreamUI --> Actions{"User Choice"}
        Actions -- "Copy" --> Clip["Write to System Clipboard"]
        Actions -- "Regenerate" --> Agent3
        Actions -- "Confirm & Paste" --> Inject["Robust React/Lexical DOM Injection"]
    end

Loading

---

📂 Repository Structure

PromptGod/
├── chrome-extension/
│   ├── manifest.json       # MV3 manifest with all_frames & scripting
│   ├── background.js       # Multi-agent orchestrator, SSE streaming, Firestore REST API
│   ├── content.js          # Shadow DOM injection, input focus tracking, robust insertion
│   ├── options.html        # Glassmorphic settings page
│   ├── options.js          # Provider & model configuration
│   ├── icons/              # 16px, 48px, 128px high-res icons
│   └── modal.*             # Standalone modal assets
├── backend/
│   ├── main.py             # Optional Python Flask microservice (Legacy / Cloud Run)
│   ├── Dockerfile          # Containerized backend deployment
│   ├── requirements.txt    # Python dependencies
│   ├── run.bat             # Windows runner
│   └── setup.bat           # Environment installer
├── web/                    # Landing page deployed on Cloudflare Pages
│   └── index.html          # Dynamic landing page showcasing features & architecture
└── README.md

---

🚀 Getting Started

1. Installation in Google Chrome

1. Clone the repository:

   git clone https://github.com/muzahirabbas/PromptGod.git
   cd PromptGod

2. Open Google Chrome and navigate to:

   chrome://extensions/
   

3. Enable Developer mode toggle in the upper right corner.
4. Click Load unpacked and select the chrome-extension folder inside this repository.
5. The PromptGod icon will appear in your Chrome toolbar. Pin it for quick access!

---

2. Configure Your AI Provider

1. Right-click the PromptGod extension icon and click Options (or click "Details" -> "Extension options").
2. Select your preferred AI Provider:
   • Google AI Studio (Recommended for lightning-fast speeds): Paste your API key from aistudio.google.com.
   • OpenRouter: Paste your free/paid key from openrouter.ai/keys.
   • Groq: Paste your ultra-fast key from console.groq.com/keys.
3. Select your model (e.g. gemini-2.5-flash-lite, google/gemma-4-31b-it:free, llama-3.3-70b-versatile, or choose Custom Model...).
4. Click Save Settings.

---

3. Usage Workflow

1. Focus or Highlight: Click inside any text box (e.g., ChatGPT, Midjourney, WhatsApp, Claude) or highlight an existing draft prompt.
2. Invoke PromptGod:
   • Press Ctrl + Shift + P (Mac: Cmd + Shift + P), OR
   • Right-click and choose Enhance with PromptGod, OR
   • Click the extension icon on your browser toolbar.
3. Review Your Draft: In the popup, your captured prompt is displayed in the Your Draft Prompt box. You can modify it or paste additional context.
4. Choose Enhancement Mode: Click one of the auto-detected functions (e.g. Image Generation, Code Optimization, Story Narrative).
5. Stream & Confirm:
   • Watch the AI stream the enhanced prompt token-by-token.
   • Click Copy to copy to clipboard, Regenerate for another take, or Confirm & Paste to inject the prompt directly into your target input!

---

🛡️ Security & Privacy

• Direct Client-to-API Communication: Your API keys are stored exclusively inside your browser's private chrome.storage.sync and sent directly to Google / OpenRouter / Groq. No private proxies or intermediary servers inspect your keys.
• Isolated Shadow DOM: PromptGod's UI exists within a closed Shadow DOM container, preventing host page scripts from reading or intercepting your workspace.
• Zero Data Harvesting: Prompts are processed strictly for real-time enhancement.

---

🤝 Contributing

Contributions, feature requests, and issues are always welcome! Feel free to open an issue or submit a pull request on GitHub.

---

📄 License

Distributed under the MIT License.