Autonomous Content Strategist Agent

This project is an intelligent autonomous agent and a full-stack web application. The backend agent automates the entire content discovery and creation pipeline, while the live web dashboard provides a user-friendly interface to monitor the agent's activity.

The project demonstrates a sophisticated Perceive -> Orient -> Decide -> Act -> Report loop, making it an excellent example of a practical, deployed agentic AI system.

Key Features

• Intelligent Backend Agent:

  • Multi-Source Monitoring: Actively monitors a list of user-defined RSS feeds for new content.
  • Content Scraping: Scrapes source articles to get a summary for deeper context.
  • LLM-Powered Relevance Filter: Uses a fast LLM call to decide if an article is "significant" enough to post about.
  • Agentic Memory: Logs all processed articles in a structured seen_articles.csv file, providing a clear audit trail.
  • Automated Content Creation: Generates multi-part Twitter/X style threads with relevant hashtags in a specific brand voice.

• Web Dashboard & UI:

  • Flask Front-End: A clean web interface for interacting with the agent's data.
  • Live & Deployed: The web application is deployed on a cloud platform (Render) and is accessible via a public URL.
  • Activity Log Dashboard: A dedicated page that displays the seen_articles.csv log in a readable table, showing all of the agent's past actions.

• The Handoff Protocol (Notification System):

  • Automated Delivery: The agent dispatches completed drafts directly to configured Slack and/or Discord channels for final human review.

• Production-Ready & Automated:

  • Cloud-Native Deployment: The backend agent runs on a schedule via GitHub Actions, and the frontend is deployed as a persistent web service.
  • Production Web Server: Uses Gunicorn for a robust, production-ready frontend deployment.
  • Rate Limit Handling: Includes an exponential backoff retry mechanism to gracefully handle API rate limits.

Demo Output

The agent delivers its work via notifications. Here is an example of a notification sent to Discord:

Discord Notification

Slack Notification

Tech Stack

• Backend: Python 3.8+, Google Gemini API
• Frontend: Flask (Python Web Framework), HTML, CSS
• Deployment: Gunicorn (WSGI Server), Render (Cloud Hosting), GitHub Actions (CI/CD)
• Core Libraries:
  • google-generativeai, feedparser, requests, beautifulsoup4, python-dotenv, Flask, gunicorn

Project Structure

/autonomous-content-agent/
|
|-- .github/workflows/run_agent.yml
|-- templates/
|   |-- landing_page.html
|   |-- dashboard.html
|-- .venv/
|
|-- .env
|-- .gitignore
|-- Procfile                       # Command for the production web server
|
|-- app.py                         # The Flask web server application
|-- agent.py                       # The backend autonomous agent script
|-- config.py
|-- brand_voice.txt
|
|-- requirements.txt
|-- seen_articles.csv

Getting Started & Deployment

1. Local Setup

First, set up and run the project on your local machine for testing.

# Clone the repository
git clone https://github.com/your-username/autonomous-content-agent.git
cd autonomous-content-agent

# Create and activate a virtual environment
python -m venv venv
# On Windows: venv\Scripts\activate
# On macOS/Linux: source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

2. Configuration

1. Create .env file: Create a .env file in the project's root folder.
2. Add your secrets: Add your API key and any optional webhook URLs.

   GEMINI_API_KEY="Your-Google-API-Key-Here"
   SLACK_WEBHOOK_URL="Your-Slack-Webhook-URL-Here"
   DISCORD_WEBHOOK_URL="Your-Discord-Webhook-URL-Here"
   

3. Customize Config: Open config.py to set your RSS feeds and brand_voice.txt to define the agent's tone.

3. Running Locally

This application has two parts. For local testing, you can run them in two separate terminals.

• Run the Backend Agent (Terminal 1):

  python agent.py

• Run the Frontend Web Server (Terminal 2):

  python app.py

  Now, visit http://127.0.0.1:5000 in your browser.

4. Deployment to the Cloud

This project is configured for a full cloud deployment.

1. Deploy the Frontend (Render):

   • Push your code to a GitHub repository.
   • Create a new Web Service on Render, connecting it to your repository.
   • Render will automatically use the Procfile to run gunicorn app:app.
   • In the Render dashboard, add your GEMINI_API_KEY and other secrets as Environment Variables.
   • Your web app will be live on a public URL.

2. Automate the Backend (GitHub Actions):

   • The workflow in .github/workflows/run_agent.yml is configured to run the agent.py script every hour.
   • It automatically commits the updated seen_articles.csv file back to the repository.
   • Render will detect this change and automatically re-deploy your web app, ensuring the dashboard is always up-to-date with the latest agent activity.