Made a thing. Julius fingerprints LLM services - point it at a target and it tells you if you're looking at Ollama, vLLM, LiteLLM, etc. Single binary, JSON output, works nicely in recon pipelines.

What it does:

• Detects 17+ services including Ollama, vLLM, LiteLLM, LocalAI, HuggingFace TGI, NVIDIA NIM
• Extracts available models from identified endpoints
• Single Go binary, no dependencies
• JSON/JSONL output for piping into other tools
• Blog: https://www.praetorian.com/blog/introducing-julius-open-source-llm-service-