A multi-agent research system that reports contradictions between sources instead of smoothing them into one answer.
Research Dossier is a multi-agent research system that queries a structured knowledge base and deliberately preserves disagreement between sources — it reports contradictions rather than flattening everything into one confident answer. You open a case by asking a question, agents investigate against the knowledge base, and a human approves the dossier before it is archived.
The author's own dev.to writeup says the app "got built almost entirely through prompting" — an iterative conversational process rather than hand-writing the UI from scratch — submitted as a Sanity Challenge entry on 2026-10-03. Declared stack: Next.js, LangGraph, Groq, Sanity.
中文导读：多智能体研报系统，保留信源间分歧而非抹平成单一答案，人工审批后归档。