Orchestrate six-agent debates to critique PRDs and code decisions.
npx skills add https://github.com/elsolal/Skillz-Claude --skill multi-mind
Please help me install this Agent Skill. Skill: multi-mind Source: https://github.com/elsolal/Skillz-Claude/tree/main/.claude/skills/multi-mind Command: npx skills add https://github.com/elsolal/Skillz-Claude --skill multi-mind
This Skill enables structured, multi-agent debates to challenge PRDs and code decisions, surfacing diverse perspectives and reducing blind spots before implementation.
To run Multi-Mind on a target document, provide the file path and mode (prd or review). Example:
High-intent search queries and answers about installing and using this skill.
To use multi-agent debate for code review and PRD validation, provide your target file path and mode, such as `review /path/to/Review.md` or `prd /path/to/PRD.md`, to trigger the analysis. The system then automatically orchestrates a five-round critique process.
A multi-agent debate process for software architecture decisions is an orchestrated workflow where six distinct AI agents analyze PRDs and code from multiple viewpoints. It surfaces diverse perspectives, documents divergences, and produces a final consensus report to reduce blind spots.
Multi-agent consensus works by enforcing a five-round process of Critique, Frictions, Targeted Debate, Convergence, and Consensus. This structured workflow prevents premature consensus, logs all critiques, and documents trade-offs to ensure robust decision-making and risk assessment.
Yes, you can automate documentation and audit trails for PRD critiques. The debate engine automatically generates a final Markdown report and saves it to the `docs/debates` directory, providing a structured audit trail that includes all divergences and trade-offs.
A limitation of using AI agents for targeted debate in code reviews is the strict enforcement against premature consensus, which requires completing all five rounds. This ensures comprehensive risk assessment but may increase processing time before reaching a final consensus.
boomNDSConvert feature specifications into executable implementation plans with tasks and acceptance criteria.
kk20300113-pngSynthesize conversation context and codebase insights into structured PRDs and publish them to the issue tracker.
ccfDocument architectural decisions with a MADR-style template and lifecycle.
d-o-hubOrchestrate three AI personas for multi-perspective code analysis.
