MCP server implementing Chain-of-Verification to reduce LLM hallucinations by drafting answers, planning fact-check questions, answering them independently, and revising.
A self-hosted MCP server that provides tools for structured reasoning, confidence calibration, and detecting recurring gaps in AI outputs, aiming to reduce workload by improving verification and attention allocation.
A multi-agent verification MCP server that uses cross-family LLM critics, NLI, and consistency checks to minimize hallucinations and false claims in LLM outputs.
MCP server for a modular RAG system that enables natural language question answering over enterprise documents with intent-aware routing, adaptive retrieval, and citation-backed responses.