| name | content-article-draft |
| description | Turns an idea, a series of notes, or raw material into an article draft (Habr / Medium / personal blog) — with a TL;DR, a thesis → arguments → example → implication structure, and an authorial voice. Use when "write an article," "shape these thoughts into a long-read," "make a Habr post," "article draft."
|
content-article-draft — Article draft
Goal: turn material into a first working draft of a 3-10k character article. Not a final version — the user polishes that themselves. The draft needs the right skeleton and tone.
Before starting, read:
~/.claude/rules/content-voice.md~/.claude/rules/content-formatting.md — the articles section
Step 1. Context
Ask:
- the topic in one sentence
- the audience: technical specialists at a certain level? Managers? Beginners?
- the platform: Habr / Medium / dev.to / personal blog / Telegraph?
- the source material: notes, experience, a project, an existing draft?
- the article's goal: teach something / share experience / break down a case / a provocative essay?
Step 2. Formulate the thesis
An article's thesis is a single claim that the author defends or illustrates. Not a topic, but a claim.
| Topic (bad) | Thesis (good) |
|---|
| "Microservices" | "Microservices are only worth the complexity when you genuinely have different release cycles or languages" |
| "Postgres vs Mongo" | "For most startups, Postgres covers 95% of the needs, and the remaining 5% don't justify a second database" |
| "My experience with k8s" | "After a year running k8s in production, I realized the main risk isn't complexity, but the blurring of SRE vs. dev responsibility" |
Show the thesis to the user and ask them to confirm or correct it. The thesis is the foundation.
Step 3. Build the skeleton
Standard structure:
# <Title>
> **TL;DR:** <thesis in 1-2 phrases + a brief summary>
## <Subheading 1: context / problem>
<Why this is interesting. A concrete real-life example.>
## <Subheading 2: argument / breakdown / solution>
<The main point. Code, diagrams, numbers.>
## <Subheading 3: nuances / counterarguments>
<Where the thesis doesn't hold. Where there are caveats.>
---
Show the skeleton (title + subheadings + a one-phrase description of each section) and ask if it works. Don't move on to writing the full text until the skeleton is agreed on.
Step 4. Write the draft section by section
Once the skeleton is confirmed — write section by section.
Applying content-voice.md:
- informal "you" / impersonal address
- authorial voice, not Wikipedia
- specifics, numbers, code
- no AI clichés
Applying content-formatting.md:
- code blocks with a language tag
- images/diagrams — alt text
- backticks for technical terms
- paragraphs — 1-4 sentences
Step 5. TL;DR at the top
For articles longer than 5000 characters — a TL;DR block at the top is mandatory. 2-3 lines: thesis + main takeaway + who will find it useful.
Step 6. Self-check
- the thesis carries through the whole article, it doesn't get diluted
- there are concrete examples (not just theory)
- there's code / diagrams / numbers where possible
- a counterargument is mentioned — otherwise the text reads like propaganda
- the reader can apply something after reading
- does the article still hold together if you remove the intro?
- does the ending make sense? (or is it just "thus")
Step 7. Show it and ask
Show the draft. Below it:
- length: X characters
- checks: ✓ thesis / ✓ specifics / ✓ counterargument / ✓ code
Ask:
- Accept as a first draft
- Fix a specific section
- Strengthen the thesis / change the angle
- Cut it down (if it ran to 12k+)
- Go deeper on a particular argument
Step 8. Save
Ask where to save it. If there's an Obsidian-vault or blog-repo convention, ask for the path. Otherwise suggest drafts/YYYY-MM-DD-<short-slug>.md next to the current working context.
What not to do
- don't produce an article without an explicit thesis — that's an essay, not an article
- don't quote random gurus for authority
- don't write "in this article we will look at" — cut it and get to the point
- don't glue a series of loosely related thoughts into one article — a series of 3 short pieces is better
- don't finalize the article without the author confirming the thesis — it's their text, not yours
