Andrew Yim
0bc24d00b9
Complete implementation of OurDigital skills with dual-platform support (Claude Desktop + Claude Code) following standardized structure. Skills created: - 01-ourdigital-brand-guide: Brand reference & style guidelines - 02-ourdigital-blog: Korean blog drafts (blog.ourdigital.org) - 03-ourdigital-journal: English essays (journal.ourdigital.org) - 04-ourdigital-research: Research prompts & workflows - 05-ourdigital-document: Notion-to-presentation pipeline - 06-ourdigital-designer: Visual/image prompt generation - 07-ourdigital-ad-manager: Ad copywriting & keyword research - 08-ourdigital-trainer: Training materials & workshop planning - 09-ourdigital-backoffice: Quotes, proposals, cost analysis - 10-ourdigital-skill-creator: Meta skill for creating new skills Features: - YAML frontmatter with "ourdigital" or "our" prefix triggers - Standardized directory structure (code/, desktop/, shared/, docs/) - Shared environment setup (_ourdigital-shared/) - Comprehensive reference documentation - Cross-skill integration support Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2.9 KiB
2.9 KiB
Skill Suitability Criteria
Evaluation criteria for determining if a need should become an OurDigital Claude Skill.
Core Criteria
1. Clear Trigger (Required)
The skill must have an unambiguous activation condition.
OurDigital Rule: Must include "ourdigital" keyword to avoid conflicts with other skills.
| Good Triggers | Bad Triggers |
|---|---|
| "ourdigital 블로그 작성" | "블로그 써줘" |
| "ourdigital research prompt" | "research this" |
| "ourdigital 스킬 만들기" | "make a skill" |
2. Focused Scope
Each skill should do 1-3 things well (Unix philosophy).
Word Limit: SKILL.md body should be 800-1,200 words.
| Good Scope | Over-scoped |
|---|---|
| Blog draft generation + SEO meta | Blog + Social + Email + Analytics |
| Visual prompt creation | Design + Development + Deployment |
3. Reusable Resources
Skill should bundle valuable resources worth maintaining.
| Resource Type | Examples |
|---|---|
| Scripts | export_to_ulysses.py, sync_notion.py |
| Templates | blog-template.md, research-plan.md |
| References | style-guide.md, api-config.md |
| Assets | brand-colors.json, prompt-library.md |
4. Domain Knowledge
Skill should encode knowledge Claude doesn't have natively.
| Good Domain Knowledge | Not Skill-Worthy |
|---|---|
| OurDigital brand voice rules | Generic writing tips |
| Ghost CMS API specifics | Standard markdown |
| Jamie clinic terminology | Common Korean |
5. Clear Boundaries
Skill should not overlap with existing skills.
Check Against:
- Other ourdigital-* skills
- Existing custom-skills (SEO, GTM, Jamie, etc.)
- Built-in Claude capabilities
Scoring Matrix
| Criterion | Weight | Score (0-2) |
|---|---|---|
| Clear trigger | 25% | |
| Focused scope | 20% | |
| Reusable resources | 20% | |
| Domain knowledge | 20% | |
| Clear boundaries | 15% |
Threshold: Score ≥ 1.2 (60%) to proceed with skill creation.
Decision Flow
Is there a clear "ourdigital" trigger?
├── No → Suggest using generic Claude or other skill
└── Yes ↓
Is scope focused (1-3 functions)?
├── No → Split into multiple skills
└── Yes ↓
Are there reusable resources?
├── No → Consider if prompt is sufficient
└── Yes ↓
Does it encode domain knowledge?
├── No → May not need a skill
└── Yes ↓
No overlap with existing skills?
├── No → Merge with existing or differentiate
└── Yes → CREATE THE SKILL
Alternatives to Skills
If criteria not met, consider:
| Alternative | When to Use |
|---|---|
| CLAUDE.md directive | Project-specific instructions |
| Prompt library | Reusable prompts without structure |
| MCP tool | API integration without workflow |
| Existing skill extension | Adding to current skill |