erdemcicek2026 / ToxiViral

What is ToxiViral?

ToxiViral is a full-stack web application that analyzes social media posts for two key metrics simultaneously: toxicity risk and viral potential. The system accepts tweet-like inputs including text, user metrics, and account metadata, then returns a combined risk-reach assessment. Built with Python's FastAPI on the backend and React with Vite on the frontend, it exposes REST endpoints for toxicity scoring and virality prediction separately or together. The toxicity model produces a 0-1 score labeled as Low, Moderate, or High Risk, while the virality predictor uses transformer embeddings and gradient boosting to estimate reach.

Why is it gaining traction?

The dual-model approach is the hook here -- most tools handle toxicity or virality in isolation, but ToxiViral combines both into a single recommendation. The API returns a Risk-Reach category that bridges the gap between "is this post safe?" and "will this post perform?" Developers interested in content moderation pipelines or social media analytics tooling find this valuable because the inference pipeline is straightforward to run locally via provided startup scripts. The UI is minimal but functional, letting you test inputs directly without writing curl commands.

Who should use this?

Content moderation teams building first-pass filters will find the toxicity scoring useful for flagging high-risk posts. Social media managers or growth-focused developers can use the virality predictions to A/B test post variations before publishing. The tool is also relevant for ML practitioners exploring practical pipelines that combine NLP classification with structured feature engineering -- the training notebooks demonstrate how the transformer embeddings and XGBoost components were built.

Verdict

At 19 stars, ToxiViral is an early-stage experiment rather than a production-ready tool. The credibility score of 0.6000000238418579% reflects this limited community validation. Documentation covers the API and architecture adequately, but there is no visible test suite or deployment guidance. Worth exploring for prototyping or learning purposes, but do not depend on it for critical workloads without significant validation first.