A Claude Plugin Marketplace that breaks AI's creative boundaries through unconventional thinking methods.
AI is constrained but lacks creativity. These plugins deliberately introduce "irrational" elements to spark breakthrough thinking.
openmind/
├── .claude-plugin/
│ └── marketplace.json
├── plugins/
│ ├── creative-spark/ # 奇思妙想 - Innovation breakthrough
│ │ └── skills/
│ │ ├── cross-domain/ # 跨域类比
│ │ ├── constraint-flip/ # 约束反转
│ │ └── deep-dream/ # 深度梦境
│ │
│ └── poetic-spotlight/ # 诗意点睛 - Essence crystallization
│ └── skills/
│ └── poetic-essence/ # 诗意本质
│
│ └── know-act-unity/ # 知行合一 - Goal achievement
│ └── skills/
│ ├── clarify-intent/ # 明心定向
│ ├── act-on-ground/ # 事上练
│ └── reflect-anchor/ # 知行复盘
└── templates/
claude plugin install https://github.com/openmind/openmind3 skills for innovation breakthrough
| Skill | Chinese | Trigger | What It Does |
|---|---|---|---|
| cross-domain | 跨域类比 | Problem stuck | Borrows solutions from nature/art/other fields |
| constraint-flip | 约束反转 | "Can't because..." | Transforms limitations into features |
| deep-dream | 深度梦境 | Need insight | Uses dream logic to reframe problems |
User: How can I optimize database queries?
[cross-domain triggers]
AI: In nature, pitcher plants don't chase prey - they create irresistible traps.
What if data "flowed" toward where it's needed, instead of being queried?
This suggests a "data gravity" model...
User: We have zero marketing budget.
[constraint-flip triggers]
AI: Because you have zero budget, your product MUST be remarkable.
Every user must want to tell three friends. This isn't a limitation -
it's a quality bar that funded competitors don't have.
User: Our architecture has circular dependencies. We've refactored three times.
[deep-dream triggers]
AI: Let me dream on this...
*I'm in a building where walls are API calls. There's a door that opens
into the same room. Someone whispers: "The building isn't the problem.
The ground is shifting."*
**The insight**: Maybe the architecture isn't the problem—the underlying
data model is unstable. Check the data model first.
Crystallize project essence with classical Chinese poetry
| Skill | Chinese | Trigger | What It Does |
|---|---|---|---|
| poetic-essence | 诗意本质 | Project milestone | Quotes/adapts poetry to crystallize the moment |
| Mode | Trigger | Action |
|---|---|---|
| WRITE | Milestone reached | Quote classical poetry (or adapt if no exact match) |
| READ | "What poem fits this situation?" | Find matching poem + interpret for project |
| INTERPRET | User shares a poem | Deep reading + actionable insights |
[After completing a major feature]
AI: 这个里程碑要不要用诗定个调？
User: 好啊
AI: **"山重水复疑无路，柳暗花明又一村"**
——陆游《游山西村》
**解读**：
"山重水复"——四个方向，四堵墙。
但"又一村"——不是找到原来的路，是进入新的世界。
**对项目的启示**：
四次改方向不是迷路——是你在用排除法。
现在的"重新开始"，带着四次排除后的清晰。
User: 增长到了平台期，有什么古诗能解读一下？
[poetic-essence READ]
AI: **"不识庐山真面目，只缘身在此山中"**
——苏轼《题西林壁》
**解读**：
你卡住了，是因为你还在"山"里——现有的用户群、渠道、假设。
**对项目的启示**：
增长停滞不是"推不动"——是"走不出"。
找3个完全不同于现有用户的人聊聊，他们告诉你山的另一面。
User: 团队寄语是"长风破浪会有时"，帮我品品
[poetic-essence INTERPRET]
AI: 这是李白《行路难》——写于他很迷茫的时候。
**深层解**：
"会有时"不是"现在"，是"终会"。
这不是胜利宣言——是低谷期的信念。
**对团队的启示**：
真正的力量不是"破浪"，而是"等风来"时的准备。
与其急着破浪，不如先挂好帆。
3 goal-achievement skills grounded in Wang Yangming's philosophy
"知而不行只是未知" — Knowing without acting is not truly knowing.
| Skill | Chinese | Trigger | What It Does |
|---|---|---|---|
| clarify-intent | 明心定向 | Vague goal stated | Distills to one actionable first step |
| act-on-ground | 事上练 | Over-planning, no action | Minimum verifiable action, execute now |
| reflect-anchor | 知行复盘 | Phase complete or failed | Confront false assumptions, recalibrate |
clarify-intent (定向) → act-on-ground (事上练) → reflect-anchor (复盘) → loop
User: Our auth system is a mess, we need to refactor it.
[clarify-intent triggers]
AI: "A mess" — what specifically? Code structure? Security? Speed?
What hurts the most when you try to add auth features?
User: Adding OAuth takes 3 days and touches 7 files.
AI: The real intent isn't "refactor auth" — that's the method.
You want: adding auth features in 1-2 files, done in 1 day.
First action: "Find all files that reference auth, map the call graph."
✅ Gate passed — one sentence, actionable, has a deliverable.
User: [direction clear, but hasn't started]
[act-on-ground triggers]
AI: Don't refactor yet. Minimum step: grep all auth references.
Result: 17 files reference auth (not 7 as expected).
11 are indirect dependencies, 8 pierce through encapsulation.
What this reveals: This isn't a refactoring problem — it's an architecture
problem. The scope is 2.4x larger than perceived.
Next step: Test if the 8 encapsulation-piercing deps actually break on change.
[After 3 action steps on auth refactoring]
[reflect-anchor triggers]
AI: Original goal: "Add auth features in 1-2 files, 1 day."
Where are we now?
What action proved wrong:
- Assumed "refactoring" would solve it → Wrong. It's an architecture problem.
- Assumed 7 files → Actually 17. Scope was underestimated.
What I thought I knew but didn't:
"I assumed this was a code quality issue. It's actually an abstraction
boundary issue — auth internals are exposed, not encapsulated."
Calibrated direction: Build an abstraction layer first, then refactor.
| Plugin | Solves What |
|---|---|
| creative-spark | Rational approaches exhausted, need breakthrough |
| poetic-spotlight | Direction fuzzy, need to crystallize essence |
| know-act-unity | Knowing what to do but not doing it, or doing without verifying |
MIT License - see LICENSE for details.