skills.rest
cnfjlhj
cnfjlhjCOMMUNITY
Explore all skills by this author →
GitHub

Human-Machine Brainstorm (人机风暴)

Coordinate multi-model requirement loops to converge specifications into a single source of truth.

425|37|Updated Mar 6, 2026
Search Tags:#codex#claude#multi-agent#opencode#ccb#requirement-alignment#spec-convergence
One-click install
npx skills add https://github.com/cnfjlhj/ai-collab-playbook --skill human-machine-brainstorm
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Human-Machine Brainstorm (人机风暴)
Source: https://github.com/cnfjlhj/ai-collab-playbook/tree/main/skills/full/human-machine-brainstorm
Command: npx skills add https://github.com/cnfjlhj/ai-collab-playbook --skill human-machine-brainstorm

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires ctx-transfer, ask, and includes scripts (resource) and references (resource) components.

What problem does it solve?

Reduce chaotic, unstructured multi-model brainstorming by providing a repeatable, auditable dispatcher loop that converges requirements into a single source-of-truth spec. The workflow enforces deterministic routing, ID-addressed Q&A, and per-round evidence export so human answers and model responses remain auditable and replayable.

Core Features & Use Cases

  • Dispatcher / Scribe / Divergent roles: Codex acts as the dispatcher, Claude Code as the canonical spec scribe, and OpenCode (Gemini) as the divergent thinker and prototype generator.
  • Deterministic routing and formats: Enforces question ID conventions (C-Q##, O-Q##, P-Q##), SHARED blocks, and ID-addressed human answers to make routing deterministic.
  • Operational tooling: Includes scripts to initialize a spec workspace, broadcast prompts to providers, route answers, and export per-provider history into .ccb/history for auditing and handoff.
  • Use cases: Converging product or engineering specs in live multi-model review sessions, structured requirement workshops, and preparing clean handoff packs for downstream models or engineering teams.

Quick Start

Paste the raw requirement into the Codex Chair, broadcast the round to claude and opencode, collect ID-addressed answers, and save the round to the .ccb/history directory.

Frequently Asked Questions about Human-Machine Brainstorm (人机风暴)

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I align requirements across multiple AI models for technical spec drafting?▼

Multi-model requirement alignment coordinates Codex, Claude Code, and OpenCode using a dispatcher, scribe, and divergent thinker to converge specifications via deterministic routing and ID-addressed Q&A.

What is the best way to run a structured multi-agent brainstorming session for product specs?▼

Structured multi-agent brainstorming uses a dispatcher loop to broadcast prompts to providers, collect prioritized questions and ASCII diagrams, and export per-round evidence to .ccb/history for auditable replay.

Can I use Claude Code as the single source of truth when converging specs with Codex and OpenCode?▼

Yes, Claude Code acts as the canonical spec scribe with exclusive edit access to .ccb/spec/overview.md, ensuring deterministic spec convergence while Codex dispatches and OpenCode generates divergent prototypes.

Do I need a specific environment setup to route questions between Codex, Claude, and OpenCode?▼

You need a running CCB environment with deterministic routing for ID-addressed answers, initialized spec workspace scripts, and per-provider history export directories to execute multi-model requirement alignment loops.

How does deterministic routing work in multi-model requirement alignment workflows?▼

Deterministic routing enforces question ID conventions like C-Q##, O-Q##, and P-Q## alongside SHARED blocks, ensuring human answers and model responses are correctly routed and remain auditable across rounds.

What are the limitations of using a dispatcher loop for multi-model spec convergence?▼

The dispatcher loop requires all participating models to operate within a CCB environment and depends on strict ID-addressed answer formats, meaning unstructured or free-form model outputs outside this protocol cannot be routed or audited.

Related Skills

Explore More Product & Management Skills→
corezoidcorezoid
58

simulator-init

Initialize and configure the Simulator.Company environment with OAuth authentication and .env storage.

Official
Intermediate
SeemSeamSeemSeam
17

ccb-self-diagnose

Diagnose CCB runtime issues by executing commands and analyzing system files.

Community
Advanced
ChegginCheggin
1

ask

Route prompts to local Claude, Codex, or Gemini CLIs and store artifacts.

Community
Intermediate
kelsi-andrewsskelsi-andrewss

spec

Generate FeatureSpec JSON and markdown specs from product descriptions.

Community
Advanced