This project contains n8n workflows designed for content moderation, specifically for User-Generated Content (UGC). The workflows handle PII (Personally Identifiable Information) detection, image censoring, and administrator notifications.

The solution is built using a combination of FastAPI as the backend and leverages various Upstage APIs for advanced AI capabilities.

The core workflows implemented in this project are:

1. UGC PII Detection: Analyzes text content to identify and flag any personally identifiable information.
2. Image Censoring: Processes images to detect and censor inappropriate content, including face detection.
3. Admin Notification: Automatically notifies administrators when content is flagged for PII or inappropriate imagery.

This project includes two main workflow models, as illustrated below.

Model 1 is a service that identifies personal information before uploading images for marketing purposes. It detects whether a face is present in the image and whether the text within the image contains any personal information, and then provides a notification.

Model 1

Model 2 has two main functions:

1. Periodic UGC Collection & Filtering: It periodically collects User-Generated Content (UGC) from social media. If the UGC contains sensitive information (e.g., faces, personal data), it is excluded from the collection. The collected UGC is then used to build a Retrieval-Augmented Generation (RAG) system using Upstage's Embedding API.

2. Chat-Based UGC Querying: An Agent AI is built using the collected UGC data and powered by Upstage's Solar Pro 2 LLM. This enables users to query the UGC information through a chat-based interface.

Model 2

Here is an overview of the project's folder structure:

ai-agent-hackathon-api/
├── app/
│ ├── main.py # Main FastAPI application
│ ├── controllers/
│ │ └── image_controller.py
│ ├── routes/
│ │ └── image_route.py
│ └── services/
│ └── image_service.py
├── n8n/ # n8n workflow files and diagrams
│ ├── Model 1.json
│ └── Model 2.json
├── .gitignore
├── pyproject.toml
├── README.md
└── requirements.txt

To run this project, you will need the following:

• API Keys for:
  • Upstage
  • X (Twitter)
• WebHook URL for Discord notifications
• Server with:
  • n8n
  • FastAPI
  • Redis

To run the backend server, use the following command:

uv run fastapi run

• n8n
• FastAPI
• OpenCV
• RAG
• Redis
• Upstage AI APIs
  • Solar Pro 2
  • Document parsing
  • Embeddings

This project was developed for the Pusan National University x Upstage AI Agent Hackathon.

• Event: Pusan National University x Upstage AI Agent Hackathon
• Date: 2025年11月12日 - 2025年11月19日
• Link: https://cse.pusan.ac.kr/cse/14651/subview.do?enc=Zm5jdDF8QEB8JTJGYmJzJTJGY3NlJTJGMjYwNSUyRjE3NDU2MzIlMkZhcnRjbFZpZXcuZG8lM0ZiYnNPcGVuV3JkU2VxJTNEJTI2aXNWaWV3TWluZSUzRGZhbHNlJTI2c3JjaENvbHVtbiUzRCUyNnBhZ2UlM0Q1JTI2c3JjaFdyZCUzRCUyNnJnc0JnbmRlU3RyJTNEJTI2YmJzQ2xTZXElM0QlMjZwYXNzd29yZCUzRCUyNnJnc0VuZGRlU3RyJTNEJTI2