‹ 首页

creativity-engine

@mark393295827 · 收录于 1 周前

Generate, validate, and output new ideas based on existing knowledge. Combines combinatorial creativity, cross-domain analogy, and minimum experiments. Use when the user wants fresh ideas, new product concepts, or creative solutions.

适合你,如果你需要从现有知识中生成和验证新想法或产品概念。

/ 下载安装
creativity-engine.skill双击,或拖进 Claude 桌面版 / Cowork,即完成安装↓ .skill↓ .zip
用别的 agent?下载 .zip 解压,把文件夹放进它的技能目录
Claude Code~/.claude/skills/(项目级 .claude/skills/)
Codex CLI~/.codex/skills/
Cursor自动读取上面两处目录
其他工具见其文档的「skills」目录;两个下载是同一份文件,只是名字不同
/ 通过 npx 安装 校验哈希
npx oh-my-skill add mark393295827/third-brain-v5-skills/creativity-engine
/ 通过 bash 安装
curl -fsSL https://oh-my-skill.com/install.sh | bash -s -- mark393295827/third-brain-v5-skills/creativity-engine
/ 已经装过?验证本机副本,不用重装
npx oh-my-skill verify mark393295827/third-brain-v5-skills/creativity-engine
安装目标可用 --agent / --scope 或 --to 明确指定;省略时只会在唯一已存在的 agent 目录上自动选择,零命中或多命中会停止并提示。content_hash 缺失或不一致均拒装。
118GitHub stars
~1.2K上下文体积 · 单文件
镜像托管

怎么用

技能原文 SKILL.md作者撰写 · MIT · 92b2e70

Creativity Engine

Generate novel ideas by combining existing knowledge across domains, then validate them through minimum experiments.

Usage Template

Prompt

Use creativity-engine on this challenge. Generate cross-domain combinations, rank them, and design the smallest test for the best idea.

Use Case

  • Producing new product ideas, content angles, experiments, or strategic options from existing knowledge.

Expected Result

  • The agent returns a ranked idea set plus one minimum experiment.

Output Example

  • 10 idea candidates, 3 finalists, 1 selected experiment, and a 7-day validation plan.

Verification Case

  • Each finalist idea states the source ingredients, target user, expected value, and a testable next step.
  • Post-ingest ideas are saved as experiment notes when they have a plausible minimum test.

Verified Effect

  • Brainstorming turns into ranked options with a smallest viable experiment for the best idea.
Success Metrics
  • Produces ranked options with source ingredients, target user, expected value, and risk for each finalist.
  • Selects one minimum experiment with a success signal, cost, and review date.
  • Saves the experiment when operating inside a configured wiki or creativity workspace.
When to Use
  • User wants "new ideas about X"
  • User is stuck on a problem and needs fresh angles
  • User wants to explore possibilities before committing
  • After a wiki ingest session — capitalize on new knowledge
Workflow
C1: Combinatorial Ideation — The Lego Building Blocks Method
"If you own one building block, you can build some cool stuff. If you get a second building block, you can build something more interesting. Get more building blocks, and very rapidly the number of things you can combine them into grows combinatorially or exponentially." — Andrew Ng

The AI Building Blocks you should know:

| Block | What It Does | Combine With | |-------|--------------|--------------| | Prompting | Basic LLM interaction | Any other block | | RAG | Retrieve and inject context | Knowledge bases, docs | | Evals | Measure quality | Any output task | | Guardrails | Safety constraints | Production systems | | Fine-tuning | Custom behavior | Domain-specific tasks | | Voice | Speech I/O | Accessibility, hands-free | | Agentic workflows | Multi-step autonomous | Complex tasks | | Tool use | External API calls | Real-world actions | | Embeddings | Semantic similarity | Search, clustering | | Vector DB | Store/retrieve embeddings | Knowledge management |

Combinatorial ideation process:

Scan the wiki for relevant concepts/entities, then combine them systematically:

[Domain A concept] + [Domain B concept] = Novel idea

Example: AI + Property Maintenance → AI vision diagnosis for apartment repairs

Generate 10-20 ideas across these categories:

  • Products — something people pay for
  • Services — something people hire done
  • Content — something people learn from
  • Automation tools — something that saves time
  • Business models — a new way to capture value

The more building blocks you know, the more combinations are possible.

C2: Cross-Domain Analogy

For each promising idea, ask:

  • What other domain solves a similar problem?
  • What pattern can I borrow?
  • What would this look like in [biology/physics/sports/gaming]?
C3: Minimum Experiment Design

For the top 3 ideas, design the cheapest test:

| Idea | Hypothesis | Minimum Test | Success Signal | Cost | |-----|-----------|-------------|---------------|------| | 1 | If X then Y | Interview 5 users | ≥3 say "I'd pay for this" | Free | | 2 | ... | Landing page + ads | ≥5% signup rate | $50 | | 3 | ... | Manual prototype | Works for 1 case | Time |

For Obsidian workflows, save the selected experiment to CREATIVITY_DIR/experiments/ with:

  • source ingredients
  • combination formula
  • target user
  • hypothesis
  • minimum test
  • success signal
  • next output
C4: Output Asset

After validation, if the idea survives:

  • Write a structured output page
  • Link to all sources and concepts
  • Set status (seed → growing)
Ideation Prompts
  1. What existing problem has a solution in another industry?
  2. If labor were free, what would I build? (Then use AI to replace the labor)
  3. What would the perfect version look like? (Work backward)
  4. What do people do manually that they hate? (Automation opportunity)
  5. What combination of wiki concepts has never been tried?
Classification Matrix

Evaluate each idea:

| Criterion | 1 (Low) | 3 (Medium) | 5 (High) | |-----------|---------|------------|----------| | Value | Nice-to-have | Saves time/money | Creates new capability | | Difficulty | Requires deep expertise | Feasible with existing tools | Weekend prototype | | Originality | Common | Unusual in this domain | Never seen before |

Quality Gates
  • [ ] ≥10 ideas generated across ≥3 categories
  • [ ] Top 3 have minimum experiment designs
  • [ ] Each idea cites its source concept(s)
  • [ ] Selected experiment is saved to CREATIVITY_DIR/experiments/ when working in an Obsidian vault
  • [ ] Validated ideas written to durable output
按 MIT 许可原样转载,未经改动 · 在 GitHub 查看 →

评论

登录即可评论;带「已验证安装」的,是发布者名下有本店的安装或持有记录。