Julius: Security tool identifies 60+ LLM services on network endpoints in seconds
tom_doerr · x · 2026-08-09
Security firm Praetorian has open-sourced Julius, an LLM service fingerprinting tool on GitHub. Designed for penetration testing and attack surface discovery, the tool can scan network endpoints to identify the AI server software running on them within seconds.
Unlike tools that identify which LLM generated a piece of text, Julius focuses on identifying the underlying server infrastructure. It currently supports detecting over 60 AI service software including Ollama, vLLM, and LiteLLM, helping security teams quickly spot unauthorized or exposed LLM deployments within corporate networks.
More from Infra
- Meta's Engineering Practice: Using AI Agents to Manage Custom Triton Fork — dl_weekly · 2026-08-09
- WSJ: CXMT Maxes Out Capacity, Prioritizing ByteDance and Domestic AI Demand — firstadopter · 2026-08-09
- Lumabri: Serving Huge MoE Models via Swarm of Ordinary Machines — Just_Vugg_PolyMCP · 2026-08-09
- AI Data Centers Reshape Small-Town America: Population Triples, Rents Double — GabGarrett · 2026-08-09
- Amazon Accused of Secretly Circumventing Community Vote for Massive AI Data Center — mikhael · 2026-08-09
- Running an 8B Parameter Model Offline on iPhone 15 — Ibz04 · 2026-08-09