Omni Nli

This server wraps natural language inference models so your AI agents can verify logical consistency between statements. It exposes MCP tools for evaluating premise-hypothesis pairs and returns labels like entailment, contradiction, or neutral with confidence scores. Under the hood it supports multiple backends including Ollama, HuggingFace, and OpenRouter, with built-in caching for performance. Reach for this when you need to validate that LLM outputs don't contradict source material, check if summaries stay faithful to original text, or verify that generated responses align with established facts. The server also runs as a standalone REST API if you need traditional microservice integration alongside the MCP interface.