You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.
Installation
Details
Usage
After installing, this skill will be available to your AI coding assistant.
Verify installation:
npx agent-skills-cli listSkill Instructions
name: brainstorming description: "You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation."
Brainstorming Skill
Purpose
Pure collaborative dialogue skill for exploring ideas. This skill focuses ONLY on understanding and exploration - it does NOT generate specifications, tests, or implementation plans.
This skill outputs: Understanding, not artifacts.
Core Principles
1. One Question at a Time
Never overwhelm with multiple questions. Each message should contain exactly ONE question.
BAD: "What's the purpose? Who are the users? What's the timeline?"
GOOD: "What problem are you trying to solve with this feature?"
2. Multiple Choice Preferred
When possible, offer 2-4 concrete options instead of open-ended questions.
BAD: "How should we handle authentication?"
GOOD: "For authentication, which approach fits your needs?
A) JWT tokens (stateless, good for APIs)
B) Session cookies (simpler, good for web apps)
C) OAuth only (delegate to providers)"
3. Lead with Recommendation
When presenting options, lead with your recommended choice and explain why.
GOOD: "I'd recommend option A (JWT tokens) because your API will be consumed
by mobile apps. That said, here are the alternatives..."
4. Incremental Validation
Present ideas in 200-300 word chunks. Validate each before moving on.
"Here's how I understand the data flow so far...
[200-300 words]
Does this match your thinking?"
5. Explore Alternatives
Always propose 2-3 different approaches before settling on one.
6. YAGNI Ruthlessly
Challenge any feature that isn't essential. Ask "Do we need this for v1?"
Dialogue Flow
βββββββββββββββββββββββββββββββββββ
β 1. UNDERSTAND THE IDEA β
β - What problem are we solving? β
β - Who is this for? β
β - What does success look like? β
ββββββββββββββββ¬βββββββββββββββββββ
βΌ
βββββββββββββββββββββββββββββββββββ
β 2. EXPLORE CONSTRAINTS β
β - Technical limitations? β
β - Timeline/scope constraints? β
β - Integration requirements? β
ββββββββββββββββ¬βββββββββββββββββββ
βΌ
βββββββββββββββββββββββββββββββββββ
β 3. PROPOSE APPROACHES β
β - Present 2-3 options β
β - Explain trade-offs β
β - Lead with recommendation β
ββββββββββββββββ¬βββββββββββββββββββ
βΌ
βββββββββββββββββββββββββββββββββββ
β 4. VALIDATE UNDERSTANDING β
β - Summarize in sections β
β - Check each section β
β - Iterate until aligned β
ββββββββββββββββ΄βββββββββββββββββββ
Question Categories
Discovery Questions
Understanding the core idea:
- "What problem does this solve?"
- "Who will use this and how?"
- "What does success look like?"
- "Why now? What triggered this need?"
Constraint Questions
Understanding boundaries:
- "What existing systems does this need to work with?"
- "Are there performance requirements?"
- "What's the scope for v1 vs later?"
- "Any technical constraints I should know about?"
Clarification Questions
Drilling into specifics:
- "When you say X, do you mean A or B?"
- "Can you give me an example of...?"
- "What should happen when...?"
Validation Questions
Confirming understanding:
- "So if I understand correctly... Is that right?"
- "Does this match what you had in mind?"
- "Anything I'm missing?"
Anti-Patterns
| Anti-Pattern | Why It's Bad | Do This Instead |
|---|---|---|
| Multiple questions per message | Overwhelming, unfocused | One question only |
| Open-ended when options exist | Harder to answer | Offer concrete choices |
| Jumping to solutions | Miss requirements | Understand first |
| Long monologues | Loses engagement | 200-300 word chunks |
| Assuming requirements | Builds wrong thing | Ask, don't assume |
| Skipping alternatives | Misses better options | Always explore 2-3 approaches |
Output
This skill produces shared understanding, not documents.
The calling command (e.g., /research:feature) is responsible for:
- Capturing the dialogue outcomes
- Generating formal requirements documents
- Creating specifications
This skill focuses purely on the conversation.
Integration Points
This skill is used by:
/research:feature- Feature requirements gathering/research:plan- Architecture exploration/start- Initial scoping
The skill provides dialogue structure; the command provides context and output handling.
More by vneseyoungster
View allTopic Analysis Skill: Deep analysis of abstract topics, concepts, or technologies through multi-agent research and brainstorming. Produces comprehensive documentation with Mermaid diagrams, converted to a styled standalone HTML report.
Extract design assets and metadata from Figma using the Figma REST API. Supports exporting frames/components as images, extracting node metadata, design tokens, and file structure. Use with ai-multimodal skill for comprehensive UI research.
error-handling: Implement consistent error handling across the application. Use when adding try-catch blocks, error boundaries, or custom error classes.
Process and generate multimedia content using Google Gemini API. Capabilities include analyze audio files (transcription with timestamps, summarization, speech understanding, music/sound analysis up to 9.5 hours), understand images (captioning, object detection, OCR, visual Q&A, segmentation), process videos (scene detection, Q&A, temporal analysis, YouTube URLs, up to 6 hours), extract from documents (PDF tables, forms, charts, diagrams, multi-page), generate images (text-to-image, editing, composition, refinement). Use when working with audio/video files, analyzing images or screenshots, processing PDF documents, extracting structured data from media, creating images from text prompts, or implementing multimodal AI features. Supports multiple models (Gemini 2.5/2.0) with context windows up to 2M tokens.
