npx skills add https://github.com/designnotdrum/brain-jar --skill capture在对话过程中将复杂想法捕捉为图表。图表在不同会话间持续保存。
在以下情况触发视觉捕捉:
根据捕捉内容匹配图表类型:
| 情境 | 图表类型 |
|---|---|
| 头脑风暴、探索想法 | mindmap |
| 包含决策点的流程 | flowchart |
| 随时间推移的交互 | sequence |
| 系统组件与连接 | architecture(使用流程图语法) |
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
| 数据模型与关系 | erd |
| 对象关系 | classDiagram |
| 状态转换 | stateDiagram |
| 项目时间线 | gantt |
思维导图(用于头脑风暴):
mindmap
root((Central Idea))
Branch 1
Leaf 1a
Leaf 1b
Branch 2
Leaf 2a
流程图(用于流程):
flowchart TD
A[Start] --> B{Decision}
B -->|Yes| C[Action 1]
B -->|No| D[Action 2]
C --> E[End]
D --> E
序列图(用于交互):
sequenceDiagram
participant User
participant API
participant DB
User->>API: Request
API->>DB: Query
DB-->>API: Result
API-->>User: Response
ER 图(用于数据模型):
erDiagram
USER ||--o{ ORDER : places
ORDER ||--|{ LINE_ITEM : contains
PRODUCT ||--o{ LINE_ITEM : "ordered in"
创建图表时始终提供丰富的上下文:
不良上下文:
"架构图"
良好上下文:
"实时通知系统的架构。展示事件如何从用户操作通过事件总线流向连接的客户端。在头脑风暴移动应用项目的通知功能时创建。"
上下文有助于您(和 Claude)在后续回顾时理解图表的用途。
对于包含大量上下文的复杂对话,可选择使用 Haiku 子代理进行草稿:
[Task tool with model: "haiku"]
Prompt: "Review the conversation above about [topic]. Draft a [diagram-type]
diagram capturing the key concepts discussed. Return only valid Mermaid syntax."
适用场景:
跳过子代理的情况:
草稿仅是起点 — 保存前请审查并完善。
图表捕捉后可以:
get_diagram 和 update_diagram 检索和修改search_diagrams 按内容或上下文搜索export_diagram 导出为 Mermaid 文件或其他格式每周安装数
1
代码仓库
GitHub 星标数
2
首次出现
今日
安全审计
安装于
zencoder1
amp1
cline1
openclaw1
opencode1
cursor1
Capture complex ideas as diagrams during conversations. Diagrams persist across sessions.
Trigger visual capture when:
Match the diagram type to what you're capturing:
| Situation | Diagram Type |
|---|---|
| Brainstorming, exploring ideas | mindmap |
| Process with decision points | flowchart |
| Interactions over time | sequence |
| System components and connections | architecture (use flowchart syntax) |
| Data models and relationships | erd |
| Object relationships | classDiagram |
| State transitions | stateDiagram |
| Project timeline | gantt |
Mindmap (for brainstorming):
mindmap
root((Central Idea))
Branch 1
Leaf 1a
Leaf 1b
Branch 2
Leaf 2a
Flowchart (for processes):
flowchart TD
A[Start] --> B{Decision}
B -->|Yes| C[Action 1]
B -->|No| D[Action 2]
C --> E[End]
D --> E
Sequence (for interactions):
sequenceDiagram
participant User
participant API
participant DB
User->>API: Request
API->>DB: Query
DB-->>API: Result
API-->>User: Response
ERD (for data models):
erDiagram
USER ||--o{ ORDER : places
ORDER ||--|{ LINE_ITEM : contains
PRODUCT ||--o{ LINE_ITEM : "ordered in"
Always provide rich context when creating diagrams:
Bad context:
"Architecture diagram"
Good context:
"Architecture for the real-time notification system. Shows how events flow from user actions through the event bus to connected clients. Created while brainstorming the notification feature for the mobile app project."
Context helps you (and Claude) understand the diagram's purpose when you return to it later.
For complex conversations with lots of context, optionally use a Haiku subagent to draft:
[Task tool with model: "haiku"]
Prompt: "Review the conversation above about [topic]. Draft a [diagram-type]
diagram capturing the key concepts discussed. Return only valid Mermaid syntax."
When this helps:
Skip the subagent when:
The draft is a starting point—review and refine before saving.
Once captured, diagrams can be:
get_diagram and update_diagramsearch_diagramsexport_diagramWeekly Installs
1
Repository
GitHub Stars
2
First Seen
Today
Security Audits
Gen Agent Trust HubPassSocketPassSnykPass
Installed on
zencoder1
amp1
cline1
openclaw1
opencode1
cursor1
统计分析技能指南:描述性统计、趋势分析与异常值检测方法
867 周安装