Claude Skills Explained — and How We Turn One Draft Into Three Platform-Tuned Posts

Cross-platform posting is a 45-minute tax per idea — three different audiences, three sets of rules, three rewrites. We codified our workflow into a Claude Skill that turns one draft into three platform-tuned posts in 30 seconds.

What are Claude Skills?

Anthropic released Claude Skills as a way to give Claude persistent specialist behavior without you re-prompting every time.

The mechanic is dead simple:

1. You create a folder like .claude/skills/my-skill/
2. Inside, a single file SKILL.md with YAML frontmatter and markdown instructions
3. Claude reads ALL available skills' frontmatter at session start
4. When you ask something that matches a Skill's description field, Claude applies it automatically

No slash command. No /skill use. No setup. The Skill is just there when you need it.

This is fundamentally different from prompts you copy-paste each time. Prompts are ephemeral; Skills are persistent specialists Claude knows about across every session in that project.

Why Skills are better than prompt libraries

Most teams accumulate folders of "good prompts" — linkedin-post-prompt.md, pricing-email-prompt.md, etc. The problem: you have to remember they exist and copy-paste each one.

Skills fix this with three changes:

1. Discoverability — Claude reads frontmatter automatically. You write naturally, the Skill activates.
2. Versioning — they live in your project's git repo. Team changes track. No more Slack-threads of "the latest version of the prompt".
3. Composability — Claude can apply multiple Skills in one response. A pricing email might use both a pricing-rules Skill and a cold-email-voice Skill.

For solo builders this saves minutes per day. For teams, it makes prompt-craft a shipping artifact rather than tribal knowledge.

Why we built necl-content-poster

We run a Telegram channel (@NeCL_company), a LinkedIn personal profile, a Threads account, and a Twitter/X account. Each has different rules:

• Telegram (RU): warm, founder-shares-cool-stuff. Brand mentions OK.
• LinkedIn (EN): decision-makers reading on mobile. Anti-hype voice. 1200-1800 characters.
• Threads (EN): one inference, screenshot-bait, 15-50 words.

Writing three posts manually from one idea is 45 minutes of tone-tuning. We did the math — we were posting maybe twice a week because of that friction.

We codified our cross-posting workflow into a Skill. Now: one draft → all three posts in 30 seconds.

What's inside the Skill

The Skill encodes three things we figured out the hard way:

1. Hook archetypes (8 of them)

LinkedIn algorithm rewards posts that pass the "see more" cut. To pass that cut, the first 1-2 lines must stop a scroller. We codified 8 hook patterns that work:

• Shock-stat: "After 4 years with ChatGPT, it still doesn't know who I am."
• Counter-narrative: "Speed is no longer your competitive advantage."
• Confession: "I used to do 10-hour days. The result? Full burnout."
• Provocation: "Most 'AI productivity hacks' are just theater."
• Dated story: "Last Tuesday, our agent stack burned $400 in 90 minutes."
• Question: "What if cache hit rate is now a moat?"
• List teaser: "5 things I stopped doing after we hit $50K MRR."
• Comparison: "I tried 3 frameworks. Only one survived production."

The Skill makes Claude pick ONE archetype and commit to it. No mushy mixed hooks.

2. Banned phrases (anti-AI-marker rules)

LinkedIn audience identifies GPT-text in two seconds. Banned in our Skill:

• "In an era where…"
• "Let's dive in"
• "It's not just X — it's Y"
• "navigating the landscape"
• "leverage", "unlock", "unleash", "game-changer", "paradigm shift"
• "at the intersection of"

If a draft would have any of these, the Skill rewrites the line. We learned this list by reading hundreds of GPT-written LinkedIn posts and noting which patterns repelled engagement.

3. Per-platform structure rules

LinkedIn: Hook → Context → Twist → Useful Core → Close + specific question → optional PS. The Twist (the line that flips the obvious reading) is mandatory. Skipping it produces a blog post, not a LinkedIn post.

Threads: ONE inference, 15-50 words, screenshot-bait or argue-bait or nod-bait. No padding. If you can split into two posts, you cut harder.

Telegram (RU): warm, founder-tone, emojis as semantic markers (3-7 per post), light brand mentions OK, soft CTA in question form at the end.

Install in 30 seconds

The fastest path uses the skills.sh CLI — works with Claude Code, OpenCode, Codex, Cursor, and 70+ other agents:

# project-level
npx skills add adjacentai/necl-skills

# or global (available in any project)
npx skills add adjacentai/necl-skills -g

That's it. The CLI drops the skill into the right folder for your agent automatically. Then in Claude Code:

What Skills are available?

You should see necl-content-poster in the list. Now just describe what you want:

Write a post about our new RAG system that cut due-diligence from 2 months to 15 minutes.

You'll get TG/LinkedIn/Threads versions in one shot.

Plug in your company context

Before first use, create a company-context.md in your project root with:

• Brand name + tagline
• Audience per platform
• Tone-of-voice rules
• CTAs to use (Calendly, DM, website)
• Reference accounts whose voice you admire
• 2-5 of your own best-performing posts (the Skill will mirror their patterns)

We ship a template — fill in 10 fields, save, done. Without context, posts are generic. With context, they sound like you.

Real example output

Input:

Just shipped a RAG system for an investment fund. Due-diligence reviews went from 2 months per deal to 15 minutes. 258 deals processed in the first quarter.

Telegram (RU) output:

📊 За первый квартал — 258 сделок прошло через нашу RAG-систему. Раньше каждая сделка = 2 месяца ручного due-diligence в фонде. Теперь — 15 минут. ✅ Загружают пачку PDF → агент находит риски, инвест-тезис, цифры. ✅ Аналитик читает summary, лезет только в спорные места. Если у вас тонна документов и одни и те же вопросы — пишите в комментарии, разберём.

LinkedIn (EN) output (excerpt):

Last quarter, we shipped a RAG system that turned a 2-month due-diligence process into 15 minutes per deal. 258 deals went through it in Q1. The lesson isn't "AI is fast." Every team knows that by now. The lesson is that the bottleneck in most knowledge work isn't thinking — it's the retrieval layer between the question and the source paragraph. [...]

Threads (EN) output:

Replaced 2 months of analyst work with 15 minutes of retrieval. The bottleneck in knowledge work was never thinking. It was finding the paragraph that has the answer.

Three posts, one input, 30 seconds. Each anchored to the same fact (258 deals, 2 months → 15 min), but pitched to its platform.

Skills vs Prompts vs MCPs — when to use which

Most teams underuse Skills. Prompts get copy-pasted, MCPs handle infrastructure — but the persistent-expertise layer in between gets neglected. That's where the daily friction lives.

See also

• necl-content-poster Skill on GitHub — install + source
• Our publishing playbook — how to ship your own Skill to every marketplace
• necl-hn-mcp — pair with this for full HN → posts pipeline
• Stop using GPT-4 for everything

FAQ