The invisible AI agent living in your clipboard.

Stop Alt-Tab-ing. Just copy, wait, and paste code.

---

⚡ Why PaperClip?

Coding flows shouldn't be broken by browser tabs. PaperClip brings the power of LLMs directly to your clipboard. It runs silently in the background, watches for your commands, and swaps your prompt with code-only answers.

• Zero UI: Acts as a ghost in your shell.
• Privacy First: No clipboard history stored.
• Resilient: Automatic failover if an API key dies.

🤖 How It Works

It's magic. No, really.

1. Copy a command: agent.prompt write a quick sort in python
2. Wait a second... (The agent talks to the AI) ⏳
3. Paste! The text in your clipboard is now the code you asked for.

graph TD;
    A[User Copies Prompt] -->|Clipboard| B(PaperClip Agent);
    B -->|API Request| C{AI Provider};
    C -->|Response| B;
    B -->|Minimalize Code| D[Update Clipboard];
    D -->|Paste| E[User Code Editor];

Loading

🌟 Feature Highlights

Feature	Description
Silent Agent 👻	Runs in the terminal, works with ANY app where you can copy/paste.
Smart Mode 🧠	agent.prompt for one-off tasks, agent.promptall for hands-free continuous prompting.
Failover 🛡️	Define multiple API keys. If one fails, it tries the next one instantly.
Clean Outputs 🧹	Automatically strips "Sure, here is the code..." fluff. You get just the code.
JSON Storage 💾	Check ./sessions for full logs if you need them. No database required.

🚀 Quick Start

1. Install Dependencies

pip install aiohttp pyperclip

2. Configure Providers

Edit providers.json to add your keys (OpenRouter, local APIs, etc).

3. Run the Agent

python bot.py

That's it. You are ready to roam.

⌨️ Commands Cheat Sheet

Type these instantly into any text field and copy them to trigger the agent.

Magic Command	Action
agent.prompt <text>	Sends <text> to the AI.
agent.prompt	Activates trigger. The next thing you copy will be sent.
agent.promptall	God Mode. Every single copy is sent to AI.
agent.promptone	Disables God Mode. Back to manual triggers.
model.gem	Switch priority to Gemini models.
model.openr	Switch priority to OpenRouter models.

⚙️ Configuration

Your providers.json is the brain. Add as many models as you need:

[
  {
    "id": "primary_gpt",
    "name": "OpenRouter GPT-4o",
    "base_url": "https://api.openrouter.ai",
    "api_key": "sk-or-...",
    "model": "openai/gpt-4o-mini",
    "priority": 1,
    "enabled": true
  }
]

🗺️ Roadmap

• Basic Clipboard Interception
• Provider Failover
• Code Minimalization
• GUI Dashboard (Optional)
• Local LLM Support (Ollama)
• Voice Trigger Support

📜 License

Free to use, modify, and hack. Just don't upload your providers.json!