菜单
Guides you through building new Claude skills with a structured Q&A process. Use when you want to create a new skill from scratch, improve an existing one, or turn a repeated workflow into a reusable skill. Produces a ready-to-upload SKILL.md file.
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。
基于 SOC 职业分类
| name | skill-creator |
| description | Guides you through building new Claude skills with a structured Q&A process. Use when you want to create a new skill from scratch, improve an existing one, or turn a repeated workflow into a reusable skill. Produces a ready-to-upload SKILL.md file. |
An interactive skill-building process. You describe a workflow or pain point; I ask the right questions; together we produce a high-quality, ready-to-upload SKILL.md file.
A good skill candidate is anything you:
Answer these questions (or let me ask them one at a time):
Produce a SKILL.md with:
Test the draft by describing 2-3 real scenarios you'd use it for. Refine until the description triggers reliably and the instructions produce the right output.
---
name: your-skill-name
description: [One sentence: what it does and when to use it. Be specific — this is what Claude reads to decide whether to activate the skill.]
---
# [Skill Name]
[1-2 sentences: what this skill is and what problem it solves]
## When to Use
[Explicit triggers and conditions]
## Instructions
[Step-by-step process or framework for Claude to follow]
## Output Format
[Structure, length, tone, any required sections]
## Examples
**Input**: [Example prompt]
**Output**: [Expected result or description of it]
## Tips
[Edge cases, common mistakes to avoid, expert notes]
The description field is the most important part — it determines when Claude activates the skill. A good description:
✅ Names the specific task: "Use when writing SQL queries that need performance optimisation" ✅ Names the trigger: "Activates when user pastes a meeting transcript" ✅ Is specific enough to avoid false positives: "for Python async code" not "for code"
❌ Too vague: "Helps with writing" ❌ Too broad: "Use for any technical task" ❌ Missing trigger: only describes what it does, not when
Before uploading:
Extracts full article text and metadata from web pages, then produces a clean structured summary with key claims, supporting evidence, author stance, and original source link. Use when researching a topic, fact-checking a claim, saving content for reference, or comparing coverage across multiple sources.
Writes clear, appropriately-toned internal communications including project status updates, company newsletters, all-hands summaries, incident reports, FAQs, onboarding docs, and team announcements. Use when writing anything meant for a colleague, team, or company-wide audience.
Identifies and qualifies high-quality leads by analyzing your product or service, searching for target companies and contacts, and providing actionable outreach strategies. Use when building a prospect list, preparing for outbound sales, researching a target account before a call, or building an ICP (Ideal Customer Profile).
Implements and evaluates software design patterns including Clean Architecture, SOLID principles, domain-driven design, and system design best practices. Use when designing new systems, refactoring existing code, evaluating architectural trade-offs, or learning design patterns through concrete examples.
Analyzes meeting transcripts to surface behavioral patterns, extract decisions, flag unresolved issues, and identify follow-up actions. Use after any meeting where you have a transcript or detailed notes — especially useful for 1:1s, team syncs, retrospectives, and customer calls.
Transforms rough ideas into fully-formed concepts through structured questioning, lateral thinking, and alternative exploration. Use when you're stuck, exploring a new idea, trying to break out of conventional thinking, or need to generate multiple strong options before committing to one.