GentWriter is an AI-powered agent designed to automatically generate various social media content based on an input article URL.

It leverages Large Language Models (LLMs) via the Google AI SDK to create tailored posts for different platforms.

• Article Fetching: Retrieves the content and title of an article from a given URL.
• SEO Meta Description Generation: Creates concise, keyword-focused meta descriptions suitable for search engines.
• X.com (Twitter) Post Generation: Crafts short, engaging tweets summarizing the article, including relevant hashtags and the article link.
• LinkedIn Post Generation: Generates professional LinkedIn posts highlighting key insights from the article, tailored for a professional audience, and includes the article link.
• Modular Agent Design: Built using the google-adk framework, organizing different functionalities into distinct agents (retrieval, SEO, X, LinkedIn) managed by a sequential workflow.
• Configurable: Allows setting the underlying LLM model via environment variables.

1. Clone the repository:

git clone https://github.com/fmind/gentwriter
cd gentwriter

2. Install dependencies:

uv sync

3. Authentication:

This project uses Google Generative AI. Ensure you have authenticated your environment for Google Cloud access. This typically involves:

• Installing the Google Cloud CLI (gcloud).
• Running gcloud auth application-default login.
• Ensure the necessary APIs (like Vertex AI) are enabled in your Google Cloud project.

The application uses environment variables for configuration:

• GENTWRITER_LLM_MODEL: Specifies the Google Generative AI model to use (defaults to gemini-2.0-flash).
• GENTWRITER_LOGGING_LEVEL: Sets the logging level (e.g., INFO, DEBUG).

• just agent-api: start an API server based on FastAPI.
• just agent-eval: evaluate the agent against an evaluation set.
• just agent-deploy: deploy the agent to Google Cloud Run.
• just agent-run: run the agent locally from the CLI.
• just agent-web: run the web UI locally.

The project includes unit tests. To run them:

1. Navigate to the project's root directory.
2. Install development dependencies
3. Run the following command:

just check

This project is licensed under the MIT License - see the LICENSE.txt file for details.