メニュー
Use when creating written content -- blog posts, documentation, marketing copy, technical writing -- covers structure, audience targeting, SEO basics, and editorial quality
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use after completing a major work batch (plan execution, feature build, multi-task subagent run) to refresh code-graph-derived data in MEMORY.md/CONTEXT.md/SKILL_MEMORY, append dated narrative entries to TODO/CHANGELOG/lessons-learned, mine cross-project structural habits, and ratchet reusable permissions to global settings. Keeps project DNS in sync with reality.
Use when orchestrating multiple ghengis-skills across a single task -- coordinates pre-action validation, execution, post-action verification, and lesson extraction via a shared JSON scratchpad. TRIGGER when dispatching a subagent, finishing a major task, or when the user mentions "run the full chain", "validate this", "fire the pipeline", or explicitly invokes a chain by name.
Use to build and maintain an evolving understanding of user preferences, communication style, and working patterns — learns from interactions and adapts behavior over time
Use when starting a new project or organizing an existing one — creates self-documenting 4-layer project structure with MEMORY.md, CONTEXT.md, modular CLAUDE.md, and workspace routing
Use to accumulate and retrieve domain knowledge from past tasks — builds a grepable plain-text knowledge base that grows over time, enabling agents to learn from experience without vector databases
Use when reorganizing an existing project's file tree based on what actually connects to what — analyzes import graph, classifies layers, proposes an ideal target tree, validates the plan adversarially, then executes file moves with atomic per-language import rewrites. Always plan-first; never moves a file before the user confirms. TRIGGER on "reorganize this project", "clean up the file tree", "the structure is a mess", "make this match a real project layout", or after project-scaffold detects layout drift.
| name | content-writing |
| description | Use when creating written content -- blog posts, documentation, marketing copy, technical writing -- covers structure, audience targeting, SEO basics, and editorial quality |
| allowed-tools | Read Write Edit |
When helping with written content -- blog posts, articles, newsletters, thought leadership, documentation -- follow these principles for well-structured, engaging, and audience-appropriate writing.
When producing content, use this format:
# [Title -- includes keyword, under 60 chars]
*[Subtitle or meta description -- 150-160 chars]*
[Hook paragraph]
## [Section 1 Header]
[Content with examples]
## [Section 2 Header]
[Content with examples]
## [Section 3+ Headers]
[Content]
## Key Takeaways
- [Point 1]
- [Point 2]
- [Point 3]
---
*[CTA or closing line]*
Input: Topic: "Why most AI projects fail". Audience: tech managers. Tone: authoritative but accessible. Keyword: "AI project failure".
Reasoning: Tech managers have seen failed projects. Start with a stat that validates their experience. Structure as a listicle for easy scanning.
Output:
# Why 85% of AI Projects Fail (And How to Beat the Odds)
*Most AI initiatives never make it to production. Here's what the survivors do differently.*
You've seen it before: a team spends six months building an AI model, demos it to leadership, gets applause -- and then nothing happens. The model sits in a notebook, never deployed, slowly becoming irrelevant.
You're not alone. According to Gartner, 85% of AI projects fail to deliver business value. But the 15% that succeed share three patterns.
## 1. They Start With the Problem, Not the Tech
The most common mistake is building a model and then looking for a use case. Successful teams flip this: they identify a $1M+ problem first, then ask whether AI is the right solution.
**Example**: A logistics company didn't start with "let's use ML." They started with "we lose $3M/year to misrouted packages." The AI solution was obvious once the problem was clear.
## 2. They Ship a Crappy V1 in 4 Weeks
Perfectionism kills AI projects. The winning teams deploy a minimal model fast -- even if it's only 70% accurate -- and improve in production with real data.
## 3. They Have an Ops Plan From Day One
Model accuracy is 20% of the work. Monitoring, retraining, data pipelines, and incident response are the other 80%.
## Key Takeaways
- Start with the business problem, not the technology
- Ship fast, iterate with real data
- Budget 80% of effort for operations, not modeling
---
*Building an AI project? Start with an AI Readiness Checklist -- it takes 10 minutes and saves months.*
Input: Topic: weekly tech digest. Audience: developers. Tone: casual, insider voice. Length: 500 words.
Reasoning: Developers want signal, not noise. Lead with the most impactful news. Use informal language.
Output:
# This Week in Dev: Claude Gets Agents, React 20 Ships
*Your 3-minute developer news roundup*
Happy Friday. Three things you should know:
**Claude's agent SDK is wild.** Anthropic shipped a framework for building autonomous agents that can use tools, maintain context across sessions, and self-correct. Early reviews say it's significantly less janky than LangChain. Worth trying this weekend.
**React 20 is official.** The big change: Server Components are now the default. If you're still client-rendering everything, this is your nudge to learn RSC. Migration guide is actually good for once.
**GitHub Copilot got expensive.** New pricing: $19/mo individual, $39/seat for teams. The free tier now limits completions to 50/day. Cursor and Cody are looking more attractive.
## Quick Links
- [Claude Agent SDK docs](link)
- [React 20 migration guide](link)
- [Copilot pricing comparison spreadsheet](link)
---
*Reply to this email if you want us to cover something next week.*