TAG
Rigorous LLM research in two layers: a multi-agent dialectic synthesis prompt + an agent-loadable research skill. Hierarchy of evidence, verbatim citation gates, no fabrication.
