There's no Google Search Console for this. No way to know if e.g. GPT-5.2 is recommending your library, where it ranks you, or whether Gemini even knows you exist. We couldn't find a tool that solved this, so we built one.

GeoStorm monitors multiple AI models on a schedule. You define search terms (e.g., "best Python async framework"), and it queries ChatGPT, Claude, Gemini, etc. via OpenRouter. It then parses the responses, tracks your brand mentions, calculates recommendation share, and alerts you when things change - a new competitor appears, your ranking drops, a model stops mentioning you.

Technically: it's a single Docker container. FastAPI backend, Astro/React frontend, SQLite, APScheduler running in-process. No Redis, no workers, no external dependencies. One OpenRouter API key gives you access to all models.

> docker run -d -p 8080:8080 -v geostorm-data:/app/data --name geostorm ghcr.io/geostorm-ai/geostorm

That gives you a working instance with 90 days of demo data so you can explore everything immediately. Add your OpenRouter key to start monitoring your own project.

MIT licensed. Would love feedback on the approach - especially from anyone who's already thinking about AI-driven discovery for their tools.

GitHub: https://github.com/geostorm-ai/geostorm Webiste: https://geostorm.ai/