Blacky AI

A local, private, and uncensored AI chat client with a hacker terminal aesthetic.

Blacky AI allows you to run powerful AI models (like Mistral, Dolphin) entirely on your own computer. No data leaves your machine. It consists of two parts:

1. The Engine: llama.cpp (The "Brain" that processes text).
2. The Client: Blacky AI (The "Face" - a cool terminal-style app you interact with).

---

Features

• 100% Private: Runs offline on your localhost.
• Uncensored: Compatible with uncensored models (no "I cannot do that").
• Hacker Console: Minimalist black & white terminal UI.
• Auto-Start: Launches the AI engine automatically when you open the app.
• Real-Time Streaming: Watch the AI type out answers character by character.
• Stop Button: Interrupt generation instantly if the AI goes off-track.

---

Prerequisites (What you need installed)

Before starting, open your terminal and run these commands to install necessary tools.

Ubuntu / Debian / Linux Mint

You need C++ compilers, CMake (build tool), and the wxWidgets library (for the GUI).

sudo apt update
sudo apt install build-essential cmake git libwxgtk3.0-gtk3-dev libcurl4-openssl-dev

---

Installation Guide

Follow these steps one by one.

1. Clone the Repository

Get the code onto your machine.

git clone https://github.com/yourusername/blacky-ai.git
cd blacky-ai

2. Set Up the AI Engine (llama.cpp)

We need to download and build the AI engine inside the engine/ folder.

# Create directory if it doesn't exist
mkdir -p engine
cd engine

# Clone llama.cpp
git clone https://github.com/ggerganov/llama.cpp
cd llama.cpp

# Build it
make

(Go back to the main folder)

cd ../..

3. Download a Model

The AI needs a "model" file (the actual brain). We recommend Dolphin Mistral (Uncensored).

1. Create a models folder:

   mkdir -p models

2. Download the model file (.gguf format) and save it inside models/.
   • Recommended Link: dolphin-2.6-mistral-7b.Q4_K_M.gguf
   • Note: This file is approx 4GB.

Your folder structure should look like this:

blacky-ai/
├── engine/
│   └── llama.cpp/
├── models/
│   └── dolphin-2.6-mistral-7b.Q4_K_M.gguf
├── client/
└── ...

4. Build the Blacky AI Client

Now compile the actual chat application.

cd client/wxblacky
mkdir -p build
cd build
cmake ..
make

---

How to Use

You don't need to touch the server. The app handles everything!

1. Run the App:

   ./wx_blacky

   (Assuming you are inside client/wxblacky/build)

2. Chat: Type your message and press Enter.

3. Stop: Click the [stop] button to interrupt the AI.

4. Exit: Just close the window. The background AI server will shut down automatically.

---

Troubleshooting

"AI SERVER FAILED TO START... [ERROR]"

• Check if you downloaded the model to the exact path blacky-ai/models/dolphin-2.6-mistral-7b.Q4_K_M.gguf.
• Make sure llama-server was built successfully in step 2.

"No response from AI"

• Ensure you have enough RAM (at least 8GB recommended).
• Check your internet connection (only needed for the initial model download, not for chatting).

Input appears twice?

• This was a known bug in older versions. Pull the latest code and rebuild.

---

Credits

• Powered by llama.cpp
• Built with wxWidgets