technical-blog-writer

name	technical-blog-writer
description	Use when writing a technical blog post about something built, fixed, or learned. Triggers on: "기술 블로그 써줘", "개발 블로그", "technical blog", "글 쓰고 싶은데 어떻게 시작해?", 이 경험 블로그로 쓰고 싶어", "engineering post", "회고 글", "튜토리얼 써줘", "개발 경험 공유".
scenarios	["DB 장애 대응 경험 블로그로 쓰고 싶어","Kafka 도입 과정을 기술 블로그로 써줘","I want to write about how we cut latency by 40ms","React 마이그레이션 경험 공유하고 싶어","이 기술 결정 이유를 포스트로 정리해줘","Tutorial for the library I built"]
compatibility	{"optional":["think-tool"],"remote_mcp_note":"think-tool이 있으면 작성 전에 블로그의 핵심 스토리와 각도를 명확히 정의할 수 있습니다. Claude 설정 → MCP Servers에서 remote SSE 엔드포인트를 추가하세요."}

Technical Blog Writer

Why Technical Blogs Fail

Most technical posts never get read past the third paragraph. The cause is almost always structural, not technical: the post opens with context that the reader did not ask for, buries the interesting part, and ends without a reason to care.

Developer readers are impatient and skeptical. They arrive with a problem or a curiosity. If the first two paragraphs do not signal "this post is relevant to your situation," they leave. Good technical writing earns attention continuously — it does not assume it.

---

Phase 0 — Extract the Core Story

Before writing anything, identify the one thing the post is really about:

1. What did you build, fix, or discover?
2. What was surprising, hard, or counterintuitive about it?
3. What would a reader do differently after reading this?

If all three answers are clear, there is a post. If the answer to #3 is "nothing specific," the post needs a sharper angle.

Ask the user these questions if the topic is still vague. Do not start drafting until there is a concrete answer to all three.

---

Phase 1 — Outline

Every technical post follows this arc. The details change, but the arc does not.

1. Hook (1–3 paragraphs)

Opens with the situation, the problem, or the surprising discovery — not with "In this post, I will explain..."

Good hooks:

• A concrete moment: "We had 40ms P99 latency and needed 8ms. We had three days."
• A counterintuitive claim: "The bottleneck was not the database — it was the ORM."
• A relatable frustration: "Every Kubernetes tutorial skips the part where it stops working."

Bad hooks:

• Background history of the technology
• "First, let me introduce..."
• Definitions

2. Problem in Depth (1–2 sections)

Describe the situation with enough specificity that a reader facing something similar recognizes themselves. Include:

• What the system looked like before
• What broke, failed, or was missing
• Why naive solutions did not work

This section builds credibility. Readers trust writers who demonstrate they actually encountered the problem rather than synthesized it.

3. The Solution (core of the post)

Walk through the solution in the order you discovered it, not in the order of a clean explainer. The journey matters because it shows reasoning, not just conclusions.

Include:

• The key insight that unlocked the solution
• Code examples that are minimal and runnable (no placeholder logic)
• Diagrams described clearly if no image is available ("The request flows from the gateway → auth service → downstream handlers, each adding a span to the trace")
• Explicit tradeoffs: what this approach does not handle, and why that was acceptable

4. Results / Validation

Concrete outcomes. Numbers where possible. What changed for users, for the team, or for the codebase.

Avoid vague claims:

• Weak: "Performance improved significantly"
• Strong: "P99 dropped from 40ms to 7ms under 3k RPS load"

5. What You Would Do Differently

One paragraph on the lesson, written for the reader — not a recap of what you did, but the transferable insight.

6. Conclusion + Call to Action

End with one clear next step for the reader:

• A related resource to read next
• A question to prompt comments or discussion
• An invitation to try something ("If you hit the same issue, the diff is here: [link]")

---

Phase 2 — Draft

Write the full post in the user's voice. Principles:

Conversational but precise. Use first person. Write short sentences when explaining concepts. Reserve long sentences for narrative. Avoid the passive voice unless necessary.

Show personality. A voice makes a post memorable. Dry technical writing is harder to finish. This does not mean jokes — it means a writer who sounds like a person rather than documentation.

Avoid jargon walls. When you introduce a technical term, define it in one clause the first time it appears. After that, use it freely.

Code examples must be runnable or obviously so. Pseudocode that cannot be adapted into real code is not a code example — it is a diagram without a diagram.

Headings should be scannable. A developer reading for the third time should be able to jump directly to the section they need using headings alone. Use specific headings, not generic ones:

• Weak: "The Solution"
• Strong: "Why Connection Pooling Fixed the Latency Spike"

---

Phase 3 — Polish

Work through this checklist before the post is finished:

• The first sentence is interesting enough to earn the second sentence
• Every section heading tells the reader what they will get from reading it
• Code blocks have language tags and are consistent in style
• Numbers are present wherever a claim is quantitative
• Every technical term is defined on first use
• The conclusion answers "so what?" from the reader's perspective
• The post is under 2,000 words (unless the complexity genuinely demands more)

---

SEO-Friendly Headings

For posts intended for search traffic, the H1 and H2 headings should contain the specific terms a developer would actually search. This is not keyword stuffing — it is using the language your reader uses.

• Weak H1: "A Journey into Performance"
• Strong H1: "How We Cut API Latency from 40ms to 7ms with Connection Pooling"

Put the core keywords near the front of the heading. Search engines and skimmers both read left-to-right.

---

Length Guide

Topic type	Target length
Quick tip or single concept	400–700 words
Full problem/solution narrative	1,000–1,800 words
Deep dive or tutorial	2,000–3,500 words
Series part (multi-part)	1,000–1,500 words per part

Longer is not deeper. Every paragraph should earn its place by adding information the reader does not already have.

---

Common Mistakes

Opening with motivation, not hook. "I've been interested in distributed systems for a while" is not a hook. The reader's interest in you is proportional to what you have just given them — which at the opening is zero. Start with the situation.

Explaining the solution without the problem. Readers need to feel the problem before the solution is interesting. Spend real time on what broke and why it was hard before explaining the fix.

Skipping the tradeoffs. A post that presents only the happy path reads like documentation, not experience. Tradeoffs are where the real insight lives.

Code blocks that are too long. A 150-line code block in a blog post is unreadable. Show the key 10–20 lines with a comment pointing to the full version elsewhere.

No call to action. Readers who finish a post and have nowhere to go produce no engagement and no memory. End with something actionable.