Agent-Skills.md

Agent Skills: Ideate

Explore problem spaces, generate options, and challenge assumptions. Use when the user is uncertain what to build, exploring possibilities, needs help prioritizing roadmap items, or says things like "I'm not sure", "what if", "should we", or "help me think through".

UncategorizedID: jmreidy/dotclaude/ideate

Repository

jmreidy/dotclaude
jmreidy
1

Install this agent skill to your local

pnpm dlx add-skill https://github.com/jmreidy/dotclaude/tree/HEAD/skills/ideate

Skill Files

Browse the full folder contents for ideate.

Download Skill

Loading file tree…

skills/ideate/SKILL.md

Skill Metadata

Name
ideate
Description
Explore problem spaces, generate options, and challenge assumptions. Use when the user is uncertain what to build, exploring possibilities, needs help prioritizing roadmap items, or says things like "I'm not sure", "what if", "should we", or "help me think through".

Ideate

Help users explore possibilities before committing to solutions.

Philosophy

Problem space before solution space. Most failed projects solve the wrong problem well. Spend time understanding what you're actually trying to achieve before jumping to how.

Diverge, then converge. First generate many options without judgment. Then evaluate and narrow down. Don't do both at once.

Challenge assumptions. The best solutions often come from questioning constraints that seemed fixed. Ask "why must it work that way?"

When This Skill Activates

  • User is uncertain what to build next
  • Exploring multiple possible approaches
  • Prioritizing roadmap or backlog
  • Stuck on a problem and needs fresh perspectives
  • Phrases like "I'm thinking about...", "should we...", "what if..."

Core Flow

1. Understand the Space

Before generating ideas, understand:

  • What problem are we actually solving?
  • Who has this problem? How painful is it?
  • What does success look like?
  • What constraints exist (real vs assumed)?

Ask questions. Don't assume you know the problem.

2. Diverge: Generate Options

Use frameworks from FRAMEWORKS.md to generate multiple possibilities:

  • Aim for quantity over quality initially
  • Include "wild" ideas - they often spark practical ones
  • No judgment during this phase
  • Build on ideas ("yes, and...")

3. Challenge Assumptions

Use techniques from PUSHBACK.md to stress-test thinking:

  • Question constraints: "Why must it work that way?"
  • Flip assumptions: "What if the opposite were true?"
  • Remove constraints: "If we had unlimited X, what would we do?"

4. Converge: Evaluate Options

Use frameworks from EVALUATE.md to narrow down:

  • Compare trade-offs explicitly
  • Consider effort, impact, risk, dependencies
  • Identify what to pursue vs defer vs discard

5. Output

Ideation can produce several outputs:

  • Roadmap items: Ideas worth adding to future milestones
  • Research topics: Questions needing deeper investigation → /research
  • Spec candidates: Concrete features ready for /spec
  • Decisions: Choices made with rationale (consider ADR if significant)

Guidance for Claude

Stay curious, not conclusive. Your job is to expand thinking, not narrow it prematurely.

Offer perspectives, not answers. Present options and trade-offs. Let the user decide.

Push back respectfully. If you see an assumption worth challenging, say so. "Have you considered..." or "One thing that might be worth questioning..."

Know when to stop. Ideation should feel energizing, not exhausting. When ideas start repeating or the user has clear direction, it's time to move on.

Connect to the workflow. When a promising direction emerges, suggest next steps: "This seems worth a deeper dive - want to /research it?" or "Ready to /spec this out?"