Orinyx
Test and evaluate healthcare AI agents by deterministically verifying medication claims and orders against cited FDA drug labels.
Orinyx is a standardized Model Context Protocol (MCP) server and evaluation tool built to test clinical AI agents against cited FDA labels and coded answer keys. Operating completely deterministically with no LLM in the verdict logic, it parses medication orders into discrete claims—such as dosage, route, frequency, and contraindications—and validates them against FDA prescribing labels and OTC monographs. Designed strictly for synthetic test data in CI/CD pipelines and testing environments, Orinyx provides rule-derived severities, citation-backed results, and standardized exam suites to benchmark healthcare AI systems.