bmad-advanced-elicitation

Iterative elicitation improves weak LLM output by applying structured critique methods like socratic questioning and red teaming. It pushes the model to re-evaluate its most recent draft and generates an enhanced replacement version after user confirmation.

Section-level content refinement is handled through a doc-section replacement workflow. The process enhances the current generated section using targeted critique techniques and returns the improved version to the invoking workflow for direct replacement.

Method-driven critique loads targeted elicitation techniques from an external methods registry. It presents constrained selectable improvements in a flow order, allowing you to reshuffle options, list all methods, or proceed after refinement.

Yes, Python3 is required to run these iterative elicitation workflows. The process depends on Python scripts to load the external methods registry and orchestrate the interactive method loop for content refinement.

Party-mode compatible orchestration resolves an agent roster and persona-aware viewpoints when party-mode is active. This allows multiple agent perspectives to drive the elicitation methods during the section enhancement workflow.

Deterministic workflows for LLM critique are limited to section-level content refinement based on pre-loaded methods. They require user confirmation through constrained control options and cannot dynamically generate critique techniques outside the registry.