Show HN: Web server does not exist (github.com)

A Show HN project implements a "web server that does not exist": every HTTP response is generated on the fly by a large language model using the request path as the prompt. The repo is a lightweight Node.js (v22+) single-file experiment (index.js) that relies only on Node built-ins (no npm packages). It defaults to the Ollama API and the qwen2.5-coder:0.5b model—users must run ollama serve and pull models locally, then run node index.js to get a URL (e.g., http://localhost:8080/about-me) where any sensible text-oriented path is turned into a model-generated response. Model and endpoint settings are editable in getAPIOpt and createServer inside the file; binary file types aren’t supported. For the AI/ML community this is an intriguing demo of treating an LLM as the entire application logic/routing layer: it shows how paths can be converted into prompts to produce dynamic content, useful for prototyping generative APIs, creative demos, or studying prompt-grounding and emergent behaviors. However, the author warns it’s fragile and not production-ready—issues include nondeterministic outputs, latency and cost, prompt injection and security risks, lack of static asset handling, and inefficiency compared to conventional servers. The project is primarily educational and experimental, inspired by NeuralOS, Ch.at, and other LLM-driven system experiments.