Menu
Creates Claude Code skills with proper structure, naming, and best practices. Use when the user asks to create a new skill, wants help writing a SKILL.md, or needs to structure skill content.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
| name | creating-skills |
| description | Creates Claude Code skills with proper structure, naming, and best practices. Use when the user asks to create a new skill, wants help writing a SKILL.md, or needs to structure skill content. |
Skills extend Claude's capabilities with domain-specific knowledge and workflows.
skill-name/
├── SKILL.md # Required: metadata + instructions
├── reference.md # Optional: detailed docs (loaded on demand)
└── scripts/ # Optional: utility scripts
---
name: doing-the-thing
description: Does X and Y. Use when the user asks about Z or needs to accomplish W.
---
# Skill Title
One-line summary of what this skill does.
## Quick Start
```bash
# most common usage
```
## When to Use
- Trigger condition 1
- Trigger condition 2
## Workflow
1. First step
2. Second step
3. Verify result
| Field | Constraints |
|---|---|
name | Max 64 chars, lowercase + hyphens only, no "anthropic" or "claude" |
description | Max 1024 chars, third person, what it does + when to use it |
Naming: Use gerund form (processing-pdfs, analyzing-data) or action form (process-pdfs).
Be concise - Claude already knows general concepts. Only add what's unique to this skill.
# Bad: 150 tokens explaining PDFs
PDF (Portable Document Format) files are a common file format...
# Good: 50 tokens, actionable
Use pdfplumber for text extraction:
Set freedom appropriately:
python scripts/migrate.py --verify"Write in third person:
For multi-step tasks, provide a checklist:
## Processing Workflow
```
Progress:
- [ ] Step 1: Analyze input
- [ ] Step 2: Transform data
- [ ] Step 3: Validate output
```
**Step 1: Analyze input**
Run: `python scripts/analyze.py input.pdf`
**Step 2: Transform data**
...
Keep SKILL.md under 500 lines. Split large content:
## Advanced Features
**Form filling**: See [FORMS.md](FORMS.md)
**API reference**: See [REFERENCE.md](REFERENCE.md)
Keep references one level deep from SKILL.md.
helper, utils, toolsscripts\helper.py → use scripts/helper.pyBefore finalizing:
Cuts a new release of ai-sdk — creates a Linear issue, branches, bumps version, updates changelog, runs checks, commits, opens a PR, and returns to main. Merging the PR triggers CI to auto-tag. Use when the user wants to release a new version of the ai_sdk gem.
References Semaphore's sem-ai CLI command set, output formats, safety defaults, and legacy sem replacement patterns. Use when exact sem-ai commands or flags are needed before inspecting or mutating Semaphore state.
Debugs and triages Semaphore CI pipelines and jobs with sem-ai first, falling back to legacy sem only for interactive attach/debug shells. Use when CI is failing, flaky, stuck, queued, or needs logs, reruns, stops, or deployment-state inspection.
Find an unreserved staging namespace and promote a Semaphore pipeline into it. Use when a user asks to pick a free `revNN`/`qaNN` server, trigger `Deploy to EKS staging`, or validate that promotion/deploy completed.
Run an extremely strict maintainability review for abstraction quality, giant files, and spaghetti-condition growth. Use for a thermo-nuclear code quality review, thermonuclear review, deep code quality audit, or especially harsh maintainability review.
Promote an Amadeus PR to the shared `data` staging namespace using Semaphore `Deploy to Staging`, then verify the SLSA-backed artifact flow, deployed image, and live log contract through Kubernetes.