GitHub radar
See which AI bots crawl your website and what they find
GEOInflow is a self-hosted analytics tool that runs on your own server and shows both human visitors and AI bots crawling your site.
It's website analytics you run yourself, tracking regular visitors alongside AI crawlers (bots) from OpenAI, Anthropic, Google, and others.
When it helps
Useful if your site earns from organic or AI search and you want to see what AI bots actually find. Skip it if you don't run your own server or search traffic doesn't matter to you.
Pros
- Built-in catalog of known AI bots: OpenAI, Anthropic, Google, Perplexity, and more
- Compares your sitemap against logs to find pages bots never visit
- Distinguishes bots used for training, search, and direct user requests
- Basic bot tracking works without a paid AI model key
Cons
- Production use needs your own domain, HTTPS, and a reverse proxy
- Some documentation is only available in Chinese
- Deeper AI analysis requires a paid AI model key
How to set it up — step by step
- 1Open your AI agent (Claude Code or Codex) in your project's folder.
- 2Send the agent the text from the block below — it will deploy GEOInflow via Docker.
- 3Make sure Docker Engine or Docker Desktop with Compose V2 is running and reachable.
- 4Open http://127.0.0.1:3002 in your browser and sign in as admin/liekan.
- 5Change the default password and create your site using the connection guide.
- 6Check the AI bot visits section on the demo site to see what the reports look like.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Set up GEOInflow from https://github.com/yaojingang/GEOInflow — bring it up with the README's quick Docker script (sh scripts/liekan-deploy.sh), sign in with the initial admin/liekan login, and show me a screenshot of the AI-bot visits section on the demo site.
This is a self-hosted web service: you need Docker Engine or Docker Desktop with Docker Compose V2 — one script brings up the app, the log worker, the analytics worker, a PostgreSQL database and a Redis cache. No GPU needed. Core analytics and AI-crawler tracking work without any AI model API key — one is only needed if you turn on the optional AI analysis layer. For production, as the authors state directly, you also need your own domain, HTTPS, and a trusted reverse proxy in front of the service.
Open on GitHub▌ More finds