AugmentClaude

Brainstorming

Generate diverse ideas and debate them across multiple AI models to reach better decisions.

Installation

  1. Make sure Claude is on your device and in your terminal.

    Skills load from ~/.claude/skills/ when Claude Code starts up — so you need it on your machine first. If you don't have it yet, install it once with the command below, then run claude in any terminal to verify.

    One-time setup
    npm i -g @anthropic-ai/claude-code

    Already have it? Skip ahead.

  2. Paste into Claude Code or into your terminal.

    This copies the whole skill folder into ~/.claude/skills/brainstorming-avelikiy/ — the SKILL.md plus any scripts, reference docs, or templates the skill ships with. Works for every skill.

    What this command does
    1. Makes a new, empty temporary folder.
    2. Downloads the skill's folder, skills/brainstorming, from github.com/avelikiy/great_cto (branch main), the GitHub repository it's listed from. You get the version that is there now, not a copy we keep.
    3. Copies that folder to ~/.claude/skills/brainstorming-avelikiy/.
    4. Deletes the temporary folder.

    Installing runs none of the skill's files; it only downloads and copies them. When you later use the skill, Claude follows its SKILL.md and may run scripts that come with it. Read them first.

    How we check skills, and what we don't · Report this skill

    Faster alternative (instruction-only skills)

    Skips the clone and grabs only the SKILL.md file. Don't use this if the skill ships Python scripts, reference markdowns, or asset templates — they won't be downloaded and the skill will fail when it tries to load them.

    Quick install (SKILL.md only)
    Sign up to copy
  3. Restart Claude Code.

    Quit and reopen Claude Code (or any other agent that loads from ~/.claude/skills/). New skills are picked up on startup.

  4. Just ask Claude.

    Skills auto-activate when your request matches the skill's description — no slash command needed. Trigger phrases live in the skill's own frontmatter; you can read them in the “What this skill does” section above.

Prefer to read the source first? Open on GitHub.

When Claude uses it

Structured idea generation + multi-LLM debate for the product-owner stage. Diverge (generate genuinely different bets), debate (a 4-persona panel on 4 models argues over 2 rounds), converge (synthesize a recommendation). Used by product-owner before architect; available to architect for design-space exploration.

What this skill does

brainstorming

Idea work has three movements: diverge → debate → converge. Most teams skip the middle one and converge on the first plausible idea. The debate panel is the point of this skill.

1. Diverge — generate genuinely different bets

From a framed problem (who · cost-of-pain · why-now · success metric), produce 3–5 distinct approaches — different bets, not cosmetic variants. Force diversity along at least one axis each:

  • different user (who you serve first)
  • different wedge (the one feature you lead with)
  • different scope (concierge/manual vs full self-serve)
  • different business shape (tool you sell vs outcome you deliver)

For each: the core bet · the smallest version that tests it · the main risk. Reject near-duplicates — if two options share the same bet, drop one and push for a more contrarian alternative.

2. Debate — the idea-debate panel

Four personas, four different models, two rounds. Model diversity matters: different model families fail differently, so they catch different holes.

PersonaStance — argues…ModelInvocation
Visionarythe strongest case FOR — the 10x outcome if it worksclaude-opus-5Task, model: opus
Skepticthe strongest case AGAINST — why it fails / who tried & diedclaude-sonnet-5Task, model: sonnet
User-Advocatethe user's honest reaction — would I pay / switch / care? AND, for any product that messages or collects data from end-recipients, the recipient's consent/opt-in friction (TCPA / opt-out / spam fatigue / who refuses)claude-haiku-4-5Task, model: haiku
Pragmatistcost, time-to-ship, build-vs-buy, unit economicsKimi K2mcp__great_cto_llm_router__ask_kimi

Round 1 — opening positions (blind, parallel)

Spawn the three Task personas in one message (parallel) + call the Kimi router for the Pragmatist. Each gets the framed problem + the diverge options and only its own stance. Prompt template:

You are the {persona} on a product debate panel. Stance: {stance}. Problem: {framing}. Options on the table: {options}. Make the strongest possible {for/against} case. Be specific and concrete — name the mechanism, the comparable, the number. End with: verdict (BUILD / DON'T / PIVOT-to-which-option) + your single biggest worry.

Round 2 — rebuttal (informed)

Feed each persona the other three's Round-1 positions. Ask:

Here are the other panelists' positions: {r1_others}. Rebut the one you most disagree with. Then give your updated verdict and the one condition that would change your mind.

Convergence guard

If all four agree in Round 1, the framing was too soft — re-run with a sharper, explicitly contrarian Skeptic ("assume this is a bad idea; prove it"). Genuine consensus only counts when the Skeptic was given every chance to kill it.

3. Converge — synthesize (the chair decides)

The product-owner (Opus) is the chair, not a vote-counter. Read all eight statements and produce:

  • The decisive consideration — the one argument that actually settles it
  • Strongest FOR and strongest AGAINST (steelmanned, attributed)
  • The call — BUILD / DON'T BUILD / PIVOT, with the reason in one sentence
  • What would flip it — the kill-criteria / the condition to revisit
  • Dissent preserved — if a persona held a strong minority view, record it; don't launder it out. Tomorrow's "why didn't we think of X" lives here.

Feed this straight into the product brief's Debate digest section.

Cost note

The panel is ~$0.30–0.60 per idea (one Opus + one Sonnet + one Haiku + one Kimi call × 2 rounds). That is the cheapest insurance in the pipeline: it runs before any engineering time is spent, at the stage where "no" is free and "yes" is expensive.

Related skills