Autowritey (Beta V1) ✍️🤖

Autowritey is a content generation assistant designed to help creators, engineers, and professionals produce high-quality content faster and easier while ensuring it sounds like them.

Important

This project is currently in Beta (V1) and is actively under development. Expect frequent updates and experimental features.

🌟 Why Autowritey?

Creating content often feels like a trade-off between speed and authenticity. Most AI tools produce generic, "robot-sounding" text. Autowritey bridge this gap by using a multi-stage pipeline that prioritizes stylistic alignment and factual grounding.

🧠 How It Works: The "User Voice" Mechanism

Autowritey achieves a consistent voice through a specialized 11-stage pipeline:

1. Style Profiling: The system loads a style_profile.json that defines your voice (e.g., "practitioner-engineer"), tone constraints (e.g., "no hype"), and vocabulary preferences.
2. Few-Shot Examples: It uses real examples of your previous writing to guide the model's stylistic nuances.
3. Analysis & Fixing: Unlike simple "one-shot" generation, Autowritey generates a draft, analyzes it against your style profile, and then self-corrects violations.
4. Tone Calibration: A final pass ensures the emotional resonance of the content matches your intended audience.

🛠️ Local Setup

To run Autowritey locally, follow these steps:

1. Prerequisites

• Python 3.10+
• A machine with a GPU (recommended) or enough RAM for local LLM execution.
• SerpAPI Key for web verification.
• Hugging Face Token (with access to Gemma models).

2. Installation

# Clone the repository
git clone https://github.com/AbdulrahmanElshafie/Blog-Generator-CLI.git autowritey
cd autowritey

# Install dependencies
pip install -r requirements.txt

3. Configuration

Create a .env file in the root directory (use .env_example as a template):

SERPAPI_API_KEY=your_serpapi_key_here
HF_TOKEN=your_huggingface_token_here

4. Style Customization

Update data/style_examples.jsonl with your own writing samples and modify config/style_profile.json to define your desired voice.

🚀 Usage

Run the generator from your terminal:

python generate.py --topic "Your Topic" --notes "Bullet points of key ideas" [OPTIONS]

Parameters:

• --topic (Required): The main subject or title.
• --notes: Key points, outline, or raw thoughts.
• --audience: Target audience (default: "general").
• --search: Enable real-time web verification for factual claims.
• --platforms: Comma-separated list of social platforms (e.g., "LinkedIn,Facebook").
• --posts-per-platform: Number of variants per platform (default: 2).
• --translate-blog: Target language for blog translation (e.g., "Arabic").
• --translate-posts: Target language for social posts translation.

📄 License

This project is licensed under the MIT License.

🛠️ Roadmap

• Interface for easier style profile management.
• Support for more local and cloud-based models.
• Browser-based UI.

☕ Support the Project

If you find Autowritey helpful, consider supporting its development!

---

Created by Abdulrahman Elshafie