PortaLLM – Local AI-to-System Bridge

PortaLLM is an open-source bridge between browser-based AI chat tools (like ChatGPT, Claude, or Gemini) and your local machine. It enables intelligent conversations to trigger real-world actions — like file operations, document creation, app automation, or scripting — without API tokens, cloud lock-in, or vendor control.

Talk to your AI. Let your system respond.

---

Why PortaLLM?

Language models are powerful reasoning tools — but in the browser, they are disconnected from your actual system. PortaLLM closes this gap.

With a simple browser extension and a local server, you can:

• Use GPT to generate files or rename folders
• Trigger automation (e.g., create a Word doc, sort PDFs)
• Let LLMs guide command-line tasks
• Access your local filesystem securely and privately
• Keep everything offline, open, and under your control

---

How It Works

Browser LLM (ChatGPT, Claude)
    |
[ Extension ]
    |
Detected AI responses (text)
    |
Local MCP API (Flask server on localhost)
    |
Local actions (file ops, Office, scripts)

---

Project Structure

portallm/
├── extension/        # Chrome/Brave extension (Manifest V3, content + background scripts)
├── server/           # Flask-based local MCP endpoint
├── docs/             # Project documentation
├── examples/         # Sample JSON payloads and GPT prompts
├── tests/            # Optional test scripts
└── LICENSE           # GPL v3 license

---

Features

• Local-first: All logic runs on your machine
• Dynamic GPT-to-system connection: From conversation to automation
• No tokens. No vendor billing. No surveillance
• Modular design: Easily add more system capabilities
• Works with any browser-based LLM

---

Setup Guide

1. Start the local server (Flask)

cd server
pip install flask
python app.py

This will start a local endpoint at: http://localhost:5000/mcp-hook

---

2. Load the browser extension

1. Open chrome://extensions (Chrome or Brave)
2. Enable "Developer mode"
3. Click "Load unpacked"
4. Select the extension/ folder

The extension will detect new LLM responses and send them to your local server.

---

3. Example Use Case

ChatGPT outputs:

Please rename the following files alphabetically:
- delta.txt
- alpha.txt
- beta.txt

→ PortaLLM extension captures it → Sends it to localhost:5000/mcp-hook → Your MCP server executes the logic on your machine

---

Philosophy

This project is for developers, tinkerers, system nerds, and AI explorers who want real agency over their tools.

PortaLLM is released under the GNU GPL v3 to ensure:

• Transparency
• Open contributions
• Resistance to proprietary capture

We believe automation should be yours — not rented.

---

Contributing

We welcome developers, testers, writers, and rebels.

To get involved:

• Fork the repo
• Check out CONTRIBUTING.md
• Open issues or submit pull requests
• Share new automation modules or use cases!

---

Inspiration & Credits

• Inspired by the limitations of token-billed APIs
• Motivated by the power of Claude, ChatGPT, and open LLMs
• Influenced by the ethos of the CCC, FSF, and hacker communities

---

License

This project is licensed under the GNU General Public License v3.0 See LICENSE for details.