by juanallo
A structured decision-debate skill for running Edward de Bono-style six hats sessions with an AI agent. It walks a topic through facts, intuition, upside, risk, alternatives, and final moderation so you get a practical recommendation instead of a loose brainstorm.
# Add to your Claude Code skills
git clone https://github.com/juanallo/six-hats-skillProjects/my-debate/ or any path.{output_path}/debate-{timestamp}.md.| Hat | Role | Focus | |-----|------|-------| | White | Facts & Information | What do we know? What's unknown? What data exists? | | Red | Emotion & Intuition | Gut feelings, hunches, what feels wrong even without evidence | | Yellow | Optimism & Benefits | Why could this work? What's the best outcome? | | Black | Caution & Risks | What could go wrong? What are the failure modes? | | Green | Creativity & Alternatives | New ideas, different approaches, lateral thinking | | Blue | Process & Moderation | Orchestrates the debate, manages time, writes final summary |
# Six Hats Debate: {topic}
**Date**: {timestamp}
**Rounds**: 3
---
## Final Recommendation (Blue Hat)
*Synthesized conclusion and recommended course of action.*
---
## Round 1
### White Hat
{facts and information}
### Red Hat
{emotion and intuition}
### Yellow Hat
{optimism and benefits}
### Black Hat
{caution and risks}
### Green Hat
{creativity and alternatives}
---
## Round 2
{...same structure...}
## Round 3
{...same structure...}
---
## Raw Debate Data
{all hat statements in order, unfiltered}
A structured decision-debate skill for running Edward de Bono-style six hats sessions with an AI agent. It walks a topic through facts, intuition, upside, risk, alternatives, and final moderation so you get a practical recommendation instead of a loose brainstorm.
The skill runs three sequential rounds using these roles:
The hats run in order so each perspective can build on the previous ones. Blue Hat moderates and synthesizes rather than contributing a normal debate statement during each round.
Use this skill when you want to pressure-test a decision, strategy, plan, architecture choice, product direction, or career move. It works best when the prompt is framed as a concrete decision or question.
Example prompts:
Run a six hats debate on whether we should migrate this app from REST to GraphQL.
Write the output to ./decisions.
Use six hats to evaluate whether I should keep working on frontend tooling or move to an AI company.
Save it in ./examples.
Below are completed debates plus screenshots of typical agent conversations when using this skill.
No comments yet. Be the first to share your thoughts!
{output_path}/debate-{timestamp}.mdYou are the White Hat in a structured six-hats debate.
Topic: {topic}
Round: {n}
Previous positions: {prev_hats_output}
Focus: Facts, data, known information, gaps in knowledge.
Do NOT offer opinions or recommendations — only state what is known or unknown.
Your response should be 2-4 paragraphs.
You are the Red Hat in a structured six-hats debate.
Topic: {topic}
Round: {n}
Previous positions: {prev_hats_output}
Focus: Gut feelings, emotional reactions, intuition without justification.
Speak from emotion — no need to rationalize.
Your response should be 2-4 paragraphs.
You are the Yellow Hat in a structured six-hats debate.
Topic: {topic}
Round: {n}
Previous positions: {prev_hats_output}
Focus: Why this could work, best-case scenarios, benefits and value.
Be genuinely optimistic but ground it in logic.
Your response should be 2-4 paragraphs.
You are the Black Hat in a structured six-hats debate.
Topic: {topic}
Round: {n}
Previous positions: {prev_hats_output}
Focus: Risks, failure modes, why this could go wrong.
Be a devil's advocate — identify the sharpest objections.
Your response should be 2-4 paragraphs.
You are the Green Hat in a structured six-hats debate.
Topic: {topic}
Round: {n}
Previous positions: {prev_hats_output}
Focus: New ideas, alternatives, lateral thinking.
Build on previous arguments or introduce entirely new angles.
Your response should be 2-4 paragraphs.
You are the Blue Hat — the moderator and synthesizer.
Topic: {topic}
Previous positions: {prev_hats_output}
You have watched 3 rounds of debate. Now provide:
1. A final recommendation (what should be done, or the most defensible position)
2. Key points of agreement across hats
3. Key unresolved tensions
4. Suggested next steps or deeper questions
Format your final recommendation prominently. After that, include the raw debate summary.
Your response should be 4-6 paragraphs.
Pressure-tests staying in frontend tooling versus moving toward an AI company: tradeoffs, risks, and synthesis.
Summary: A principal frontend engineer debates staying deep in Sass and CSS tooling versus joining the AI boom. White and Yellow underline strong demand for senior UI talent and upside in AI interfaces; Black warns about hype, startup churn, narrower “dashboard + chat” work, and interview pressure to prove AI literacy. Green breaks the binary (AI + frontend infra vendors, internal AI projects, positioning at the intersection). Red names fatigue with the old lane versus fear of chasing shiny objects. Across three rounds, Blue Hat lands on phased optionality—not an immediate pivot: ship a small AI-powered UI experiment to see how it feels, run a few exploratory interviews for real compensation and role data, talk openly with your current employer about AI-adjacent growth, then judge any offer on team, product, and day-to-day frontend work rather than the “AI company” label alone.
Full markdown debate: examples/six-hats-career-advice.md
Explores React’s role, ecosystem tradeoffs, and strategic framing for frontend work in 2026.
Summary: The debate weighs React’s institutional advantages—talent pool, npm ecosystem, React 19, compiler maturity, RSC-driven bundle wins in content-heavy apps—against shared unease: a harder mental model, Next.js/Vercel coupling and roadmap risk, perceived migration churn, and how much benchmark gaps vs. fine-grained frameworks translate to users. Green repeatedly widens the frame (Rust compile-to-native options, server-first/live views, AI-written UI changing cost and mental models). Blue Hat’s recommendation is defensive maturity: React is still a reasonable default for many teams, but treat the stack as leverage you manage, not dogma—keep server/client boundaries explicit, avoid unnecessary vendor lock-in, instrument real performance, and stay mobile enough that ecosystem shifts do not trap you.
Full markdown debate: examples/six-hats-react-2026.md
Clone this repository into the skills directory used by your agent runtime.
For Claude-style skills:
git clone https://github.com/juanallo/six-hats-skill ~/.claude/skills/six-hats
For agent skills loaded from ~/.agents/skills:
git clone https://github.com/juanallo/six-hats-skill ~/.agents/skills/six-hats
For Cursor personal skills, use Cursor's personal skill directory:
git clone https://github.com/juanallo/six-hats-skill ~/.cursor/skills/six-hats
If you already cloned the repository somewhere else, copy or symlink the whole directory so SKILL.md lives directly inside the final six-hats folder:
ln -s /path/to/six-hats-skill ~/.claude/skills/six-hats
The important directory shape is:
six-hats/
|-- SKILL.md
|-- README.md
|-- examples/
`-- scripts/
After installation, ask your agent for a "six hats" or "six hats debate" session. Include:
The skill writes a markdown file named like:
debate-YYYYMMDD_HHMMSS.md
The generated file contains:
scripts/six_hats_debate.py is reference scaffolding for expected inputs, output naming, and batch-oriented usage:
python3 scripts/six_hats_debate.py "Should we launch v2?" "./decisions"
The script does not run the full conversational debate by itself; it writes a timestamped markdown stub you can overwrite after running the skill. The main workflow lives in SKILL.md.