brainstorming

Guides collaborative design exploration and spec writing before any code implementation begins.

Updated Aug 3, 2026
One-click install
npx skills add https://github.com/Yvesdefaria/GymLab --skill brainstorming-yvesdefaria
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: brainstorming
Source: https://github.com/Yvesdefaria/GymLab/tree/main/gymlab-app/.agents/skills/brainstorming
Command: npx skills add https://github.com/Yvesdefaria/GymLab --skill brainstorming-yvesdefaria

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? It prevents wasted implementation effort by forcing requirements clarification, design approval, and spec review before any code is written, catching wrong assumptions early. ## Core Features & Use Cases - Three-Path Classification: Routes every request into spike, bounded, or architectural tracks with appropriately scaled process and a mandatory human approval gate. - Spec Authoring and Review: Produces dated design documents under docs/superpowers/specs/ with a self-review checklist and a subagent reviewer prompt template. - Visual Companion Server: Ships a Node.js WebSocket server with shell launchers that serves HTML mockups and captures browser click selections during design discussions. - Use Case: Before adding a new feature to an existing app, classify the change as bounded, present a short in-chat design, get explicit approval, then implement through the normal workflow. ## Quick Start Ask the agent to brainstorm the feature you want to build so it classifies the scope, asks clarifying questions, and presents a design for your approval before writing any code.

Frequently Asked Questions about brainstorming

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

FAQPage Schema
How do I run a design brainstorming session before writing code?

Invoke the brainstorming skill before any creative work. It classifies your request as spike, bounded, or architectural, asks clarifying questions one at a time, and presents a design you must approve before implementation starts.

What is the difference between spike, bounded, and architectural tasks?

A spike answers a feasibility question with throwaway probes. Bounded covers well-scoped changes to existing code with a short in-chat design. Architectural covers new projects or subsystem restructuring and requires a written spec and implementation plan.

How do I start the visual brainstorming companion server?

Run scripts/start-server.sh with --project-dir and --open after the user approves the companion. It returns a JSON object with the keyed URL, screen directory, and state directory for pushing HTML mockups.

Does the brainstorming companion work on Windows or remote environments?

Yes. The start script auto-detects Windows-like shells and Codex environments and switches to foreground mode. For remote or containerized setups, pass --host 0.0.0.0 and --url-host to bind a reachable interface.

Why does the companion URL require a session key?

The server generates a per-session token embedded as ?key= in the URL and mirrored into a cookie. Requests without the key receive a 403, preventing other local tabs or network hosts from reading screens or injecting events.

When should I not use the visual companion during brainstorming?

Skip it for conceptual or text-based questions like requirements, tradeoffs, and scope decisions. Reserve the browser for genuinely visual content such as mockups, layout comparisons, and architecture diagrams.