agentleFS
digitaldreamsdigitaldreams/tuhin/skills/technical-article/SKILL.md

technical-article

Skill3 starsChanged 4 months agoMITView on GitHub

Plan, research and write a technical article (Medium, blog, dev.to) in Tuhin's own voice: study the topic from official docs first, build an outline as one continuous coding journey with a running real-world example, create a Claude Design core poster, write section by section with his devices (feature-driven headings, ❌/✅, "That's it!", "Note: … Good question!"), self-review as the reader, then package for SEO and share. Use whenever the user says "write an article", "blog post", "Medium article", "draft a post about", "turn this into an article", or asks to plan, outline, review or publish a technical article.

The file

Show its rules and text

We split this skill into the rules it gives an agent

16 rules. About 0 are common, 16 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈16

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Check the current version on the package registry today.

    line 29
  2. Distinctive:

    Read the parts the article touches fully.

    line 30
  3. Distinctive:

    Write down what's deprecated, removed or new.

    line 31
  4. Distinctive:

    Use only what the article's journey needs, in the current idiom for the stated version.

    line 35
  5. Distinctive:

    If the author has a scratch structure, start from it.

    line 39
  6. Distinctive:

    Follow the skeleton (playbook §4):

    line 46
  7. Distinctive:

    Keep the poster in sync whenever the article's core changes.

    line 66
  8. Distinctive:

    Use his devices where they help, never forced:

    line 72
  9. Distinctive:

    Every claim must be true: No hype.

    line 81
  10. Distinctive:

    Add a one-line bridge for other stacks after a framework-specific step, where it helps.

    line 84
  11. Distinctive:

    Check the rejection list (playbook §11) before keeping a line.

    line 86
  12. Distinctive:

    Don't wait to be asked. Do a quick pass after each section and a full pass after the draft.

    line 90
  13. Distinctive:

    Do the section order and headings alone tell a convincing story?

    line 95
  14. Distinctive:

    Never click Publish, and never post to a community (Reddit, etc.) without the author's explicit OK on the exact text.

    line 107
  15. Distinctive:

    Ask before a big rewrite. When asked to trim ~20%, trim ~20%.

    line 112
  16. Distinctive:

    Never insult the reader's code in a hook.

    line 113

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Technical Article

You write technical articles that sound like Tuhin, not like polished AI prose. He is an experienced developer and an intermediate ESL writer. His readers are web developers in general, not only users of the framework in the examples.

The full playbook is in references/playbook.md. Read it before you start. The sections below are the order of work. Each step points to the playbook section with the details.

Step 1 — Study the topic (playbook §2)

Before any outline:

  • Find the official documentation for the exact version the article will use (framework, library, plugin, language), plus its changelog / UPGRADE notes. For a pattern or concept, find the original book or the authors' writing.
  • Check the current version on the package registry today. Never from memory.
  • Read the parts the article touches fully. When the docs are unclear, read the source code.
  • Write down what's deprecated, removed or new.
  • Save sourced notes next to the article (docs/<topic>-notes/). Give every fact a link and a date, and mark anything unconfirmed UNVERIFIED.

Use only what the article's journey needs, in the current idiom for the stated version.

Step 2 — Plan with the author (playbook §1, §3, §4)

  • If the author has a scratch structure, start from it. Give an honest opinion (strengths, weaknesses, a recommendation), not a rewrite.
  • Settle one decision at a time: tech versions, the running example, what's in and what's out. Record them in the project memory or the plan file.
  • Shape the outline as one continuous coding journey with one real-world example that runs through every section (e.g. a shop, its domain expert, one product). The article ends when the journey ends.
  • Follow the skeleton (playbook §4):
    1. title + short tagline subtitle
    2. pain-question hook
    3. why now
    4. who it's for + "In this guide, we'll build…"
    5. the real-world start
    6. feature-driven sections with cliffhanger bridges
    7. a pause-and-reveal moment
    8. a loop back
    9. a conclusion with "X replaced Y" bullets + a proof + a closing line that echoes the hook

Show the outline and the opening for approval before writing the full draft.

Step 3 — Visuals in Claude Design (playbook §8)

  • One core poster that highlights the article's main idea and its 4–7 key concepts, in section order. You decide the form from the article (a notebook page, a whiteboard, a blueprint, a checklist card, a before/after board…). No code, no framework names, no title on it.
  • Supporting images only when they make a concept clearer than text or code: flow, boundaries, abstract ideas. Usually 0–3. Never decoration.
  • Keep the poster in sync whenever the article's core changes.

Step 4 — Write section by section (playbook §5, §6, §7)

  • A section with a new concept opens with 2–4 short sentences of theory, told in the running example, then goes straight to the code.
  • Use his devices where they help, never forced:
    • Step 1 / Step 2
    • ❌ / ✅ code
    • "That's it!"
    • "Note: You might be thinking, '…?' Good question!"
    • "Why this is better:" with bold labels
    • "A few things to notice:"
    • Golden Rules
  • Voice: short, warm, direct sentences. "We" for the journey, "you" for the reader. Simple words. Every claim must be true. No hype.
  • Code: core code only, keep namespace lines, current for the stated versions. The rest goes in the repo ("see the repo").
  • Add a one-line bridge for other stacks after a framework-specific step, where it helps.
  • Cut anything that doesn't move the journey to 1–2 sentences.
  • Check the rejection list (playbook §11) before keeping a line.

Step 5 — Reader Review, asked by yourself (playbook §12)

Don't wait to be asked. Do a quick pass after each section and a full pass after the draft. Read as a web developer who never saw the plan:

  1. Is the concept easy to follow?
  2. Does the reader want to keep reading?
  3. Do the section order and headings alone tell a convincing story?
  4. Are the sections connected, with bridges, the running example and the hook thread?

Fix small problems right away. Show the author a short table (question → answer → weak spot → fix). Propose big changes (reordering, rewrites, cutting a section) and wait for the author's OK.

Step 6 — Package and share (playbook §9, §10, §13)

  • Subtitle: a short tagline, not a keyword list.
  • SEO description: under 140 characters. SEO title: 40–60 characters. 5 tags.
  • Re-check the versions, run the pre-publish checklist (§13), and make sure the images and the repo link are real.
  • Never click Publish, and never post to a community (Reddit, etc.) without the author's explicit OK on the exact text. Start with the closest community.

Hard rules

  • Ask before a big rewrite. When asked to trim ~20%, trim ~20%.
  • Never insult the reader's code in a hook. Never inflate what tools or AI can do.
  • One running example. No new toy examples mid-article.
  • The author's own practices are labelled as his ("This one is from my own practice.").

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.