用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/majiayu000/claude-skill-registry --skill brainstorming-designing命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
LLM token logprobs and calibration. Per-decision confidence, ECE, Brier, reliability diagrams, low-confidence triage.
Analyze LLM token logprobs and calibration. Use for per-decision confidence, ECE, Brier scores, reliability diagrams, and low-confidence triage.
回顾最近 N 天的 Claude Code 使用记录——扫描原始会话数据,按主题分组汇总"我都做了什么",并从个人操作系统视角输出模式、风险与增删建议。当用户说 /recap、"看看我这几天做了什么"、"回顾一下我最近的会话"、"这两天我用 claude 干了啥"、"活动回顾" 时使用。
基于 SOC 职业分类
正在显示 SKILL.md
| name | brainstorming-designing |
| description | The DESIGNING phase presents the design approach in small, validated sections |
| user-invocable | false |
| allowed-tools | ["Read","Glob","Grep","Bash","AskUserQuestion","mcp__plugin_mermaid-collab_mermaid__*"] |
The DESIGNING phase presents the design approach in small, validated sections.
Before writing the design:
For each section of the design:
[PROPOSED] marker1. Accept
2. Reject
3. Edit
User responses:
[PROPOSED] marker, continue to next sectionWhen currentItem is set in collab-state.json:
Update the work item in the design doc with:
**Problem/Goal:** - documented problem/goal**Approach:** - documented approach**Success Criteria:** - documented criteria**Decisions:** - any item-specific decisionsREQUIRED for each proposed approach:
Before presenting an approach to the user, create a diagram visualizing it:
Tool: mcp__plugin_mermaid-collab_mermaid__create_diagram
Args: {
"project": "<cwd>",
"session": "<session>",
"name": "approach-N",
"content": <flowchart/sequence showing the proposed approach>
}
Do NOT describe architecture or flow in text alone. Show it visually, then explain.
When brainstorming involves visual artifacts, use the mermaid-collab server.
GUI/UI Design (ALWAYS use wireframes):
create_diagram(name, content) with wireframe syntaxpreview_diagram(id) so user can see in browserArchitecture and Flow Design:
Design Documents:
create_document(name, content) for design specsupdate_document(id, content)Workflow:
Before moving to VALIDATING, ensure:
When brainstorming within a collab session, update the design document using MCP tools.
Prefer patch operations for targeted changes:
For small, targeted edits (updating a single field, adding a bullet point, changing status):
Tool: mcp__plugin_mermaid-collab_mermaid__patch_document
Args: {
"project": "<cwd>",
"session": "<name>",
"id": "design",
"old_string": "<exact text to find>",
"new_string": "<replacement text>"
}
Use full update only when:
Fallback to full update:
Read current content: Tool: mcp__plugin_mermaid-collab_mermaid__get_document Args: { "project": "", "session": "", "id": "design" }
Modify content as needed (add sections, update decisions, etc.)
Write updated content: Tool: mcp__plugin_mermaid-collab_mermaid__update_document Args: { "project": "", "session": "", "id": "design", "content": "" }
Important: Always read before full update to preserve existing content.
Can return to CLARIFYING phase if:
Announce: "Let me go back and clarify that before continuing the design."
Prerequisites:
Announce: "Design sections complete. Let me run the completeness gate."
At the end of this skill's work, call complete_skill:
Tool: mcp__plugin_mermaid-collab_mermaid__complete_skill
Args: { "project": "<cwd>", "session": "<session>", "skill": "brainstorming-designing" }
Handle response:
action == "clear": Invoke skill: collab-clearnext_skill is not null: Invoke that skillnext_skill is null: Workflow complete