BS
Brainstorming Skills: Architecture Design & Requirement Exploration
obra
2026-04-06

When product and R&D teams use openclaw/hermes agent for new features, brainstorming skills help deeply explore user intent and architecture design, using state machines to standardize business logic and reduce rework.

Introduction

In daily development, unexamined assumptions often lead to wasted effort, even in seemingly simple tasks. The brainstorming skill tackles the pain point of “coding before clarifying” by acting as a mandatory design gate. It forces the AI to engage in a structured dialogue with the developer before writing any logic, transforming ambiguous ideas into concrete design specifications (Specs). This prevents architectural missteps and provides a clear blueprint for implementation.

Concept

Design first, implement later. Forces the AI to understand project context, clarify constraints, and generate a concrete spec before coding, replacing expensive refactoring with low-cost early communication.

Setup and Usage

There are several ways to install a skill:

  • Method 1: In the OpenClaw or Hermes Agent chat window, directly tell the Agent: Please help me install the brainstorming skill. (Easiest)
  • Method 2: Visit the skillhub website, install the skillhub store first, and then install the corresponding skill. (For Chinese users)
  • Method 3: Visit the Skills.sh website, search for the skill name on the homepage, and use the provided command to install. (For experienced users)
  • Method 4: Visit the Clawhub website, search for the skill name, download the zip file, extract it, and place it in the OpenClaw skills directory.

Skill Workflow Analysis

  • Explore Project Context: Before asking questions, the AI must check existing files, docs, and recent commits to understand the current workspace.
  • Offer Visual Companion (Optional): If the topic involves UI or layouts, the AI explicitly asks (in a standalone message) whether to launch a browser-based visual tool, decoupling visual confirmation from logic.
  • Single-Step Clarification: The AI is strictly limited to asking one question at a time. This prevents information overload and ensures precise developer feedback on constraints and success criteria.
  • Provide Multi-Dimensional Options: The AI is prohibited from running with its first intuition. It must propose 2 to 3 architectural approaches with explicit trade-offs and recommendations, helping developers evaluate alternatives.
  • Progressive Design Validation: Complex systems are broken down into smaller, manageable sections (architecture, data flow, etc.). The developer must approve each section before proceeding, creating a secure validation loop.
  • Generate and Self-Review Spec: After discussion, the AI generates a physical Markdown design document and performs an inline self-review to eliminate placeholders, “TBD”s, or ambiguous requirements.
  • Hard Gate to Planning: The final node of this workflow strictly triggers the writing-plans skill. This acts as a hard gate, completely blocking the AI from skipping the planning phase to write implementation code.

Skill Design Evaluation

SKILL.md

Back to Product Manager Skills List