These 14 blog writing prompts take a post from idea to published draft: picking a topic, outlining, writing the first draft, sharpening the intro, editing, checking facts and turning one post into several pieces. Each prompt asks for your real material first, because a model with only a topic name writes the same post as everyone else.
How to use these prompts
Every prompt works in ChatGPT, Claude or Gemini. Replace the blanks in double braces, or keep the default after the colon: {{audience:junior developers}} means "junior developers" unless you change it.
Run them in order for a new post. Outline first, then draft, then edit. Each step is a checkpoint where you can fix the direction cheaply. OpenAI's prompt engineering guide for everyday work frames a prompt as a handoff to a smart intern: the background, the specific task, and what "good" looks like in audience, tone and format. The prompts below are built the same way.
Paste your own notes, examples and opinions wherever a prompt asks for them. That material is what makes the post worth reading.
Which prompts help you choose a topic?
1. Topic ideas from reader questions
Start from what readers already ask, not from what is trending.
Here are questions my readers have asked in comments, emails and support
tickets:
{{questions}}
My blog is about {{niche:web development for small teams}} and my readers
are {{audience:junior developers}}.
Suggest 10 blog post ideas that answer these questions. For each give a
working title, the one question it answers, and why a reader would click
it over a generic article. Group ideas that could be one post.
2. Find the angle nobody else took
For a topic that already has a hundred posts. Paste the headings of the top results so the model sees what exists.
I want to write about {{topic:choosing a database for a side project}}.
Here are the headings from articles that already rank for it:
{{competitor_headings}}
What do these articles all say? What do they skip, get wrong or assume?
Suggest 3 angles I could take that add something new, based on this
experience of mine: {{experience}}
Which prompts help you outline a post?
3. Outline with one point per section
An outline is the cheapest place to fix a post. This one forces every section to earn its spot.
Write an outline for a blog post titled "{{title}}" for {{audience:junior developers}}.
The one thing the reader should take away: {{takeaway}}
Target length: {{length:1,200 words}}
Give H2 headings, and under each one sentence on the point it makes and
the example or evidence it uses. Open with a section that answers the
title directly. Cut any section that doesn't serve the takeaway.
My notes:
{{notes}}
4. Split a big topic into a series
When the outline keeps growing past what one post can hold.
I planned one post on {{topic:migrating a monolith to services}}, but the
outline is too long. Here it is:
{{outline}}
Split it into a series of {{parts:3}} posts. For each post give a title,
the one question it answers, the sections it takes from the outline, and
what a reader needs from earlier posts. Each post must be useful on its
own. Point out any section that belongs in none of them.
Which prompts write the first draft?
5. First draft from an approved outline
Paste the outline you edited, plus a sample of your writing.
Write the first draft of this blog post from the outline below.
Rules: write in the style of my sample, short paragraphs, plain words,
one idea per paragraph. Use only the examples and facts in my notes. If a
section needs a fact or example I didn't give, write [NEED EXAMPLE] or
[NEED SOURCE] instead of inventing one. Target length: {{length:1,200 words}}.
Outline:
{{outline}}
Notes:
{{notes}}
Writing sample:
{{writing_sample}}
6. Plan the examples, code and visuals
Walls of prose lose readers. This finds the spots where a snippet, screenshot or table would explain faster.
Read this draft for {{audience:junior developers}}. For each section,
say whether a code snippet, screenshot, diagram, table or before/after
example would make the point faster than the text. For each one you
suggest, describe exactly what it shows and write a one-line caption
and alt text. Don't invent code or numbers; describe what I should
capture from my own work. Suggest no more than {{max_visuals:5}} in total.
{{draft}}
7. Tutorial post with steps that work
How-to posts fail when a step is missing. This prompt asks for the prerequisites and the checks.
Write a step-by-step tutorial on how to {{task:deploy a Next.js app}} for
{{audience:junior developers}}.
Start with what the reader will have at the end and what they need
before starting. Number the steps. After each step, say how the reader
can tell it worked. End with the 3 most common errors and their fixes.
Use only the commands and settings in my notes; mark anything else
[CHECK].
Notes:
{{notes}}
Which prompts sharpen headlines and intros?
8. Headline options
Asks for variety so you can compare rather than accept the first idea.
Write 10 headline options for this blog post. Mix: a direct "how to", a
specific number, a question the reader is asking, a mistake to avoid,
and a plain statement of the takeaway. Keep each under {{max_chars:60}}
characters and show the count. Don't promise anything the post doesn't
deliver.
{{draft}}
9. Intro that gets to the point
Most AI intros spend three sentences warming up. This one answers first.
Rewrite the introduction of this post. The first sentence must answer
the question in the title or state what the reader will get. Then one or
two sentences on why it matters to {{audience:junior developers}}. No
"in today's world", no rhetorical questions, under {{words:80}} words.
Give 3 versions.
Title: {{title}}
Current intro:
{{intro}}
Which prompts edit and check the draft?
10. Line edit for clarity
A pass that tightens without rewriting your voice.
Edit this draft for clarity. Shorten long sentences, replace vague words
with specific ones, cut repetition, and fix grammar. Keep my voice and
my examples. Show the edited text, then a list of the 5 biggest changes
and why. Don't change technical terms or quotes.
{{draft}}
11. Read it as a skeptical reader
Finds the weak spots a friendly editor would let slide.
Read this post as a {{reader:senior engineer who thinks the topic is overhyped}}.
List every place where you would stop reading, roll your eyes or ask
"says who?". For each, quote the line and suggest a fix: an example, a
source, a caveat or a cut.
{{draft}}
12. Fact-check list before you publish
Models can get version numbers, flags and prices wrong. This prompt doesn't ask the model to judge its own answers: it builds the list you check by hand. Anthropic's guide to reducing hallucinations suggests giving the model explicit permission to admit uncertainty, which the last line does.
List every checkable fact in this draft: version numbers, commands,
flags, config keys, prices, limits, dates, quotes and claims about what
a product does. Return a table: the exact line, what kind of fact it is,
the official source I should open to confirm it (docs, changelog,
pricing page), and how likely it is to go out of date. Do not tell me
whether a fact is right; I will check. If you can't tell whether
something is a fact or an opinion, list it anyway and say so.
{{draft}}
Which prompts help after you publish?
13. Repurpose into other formats
One post can feed a newsletter and a week of social posts. For more on the social side, see the social media prompts.
Turn this blog post into: a newsletter section under {{newsletter_words:200}} words,
{{post_count:3}} posts for {{platform:LinkedIn}}, and a 5-point summary
for the top of the post. Each piece should stand on its own and point
back to the full post. Keep my voice.
{{post}}
14. Refresh an old post
For posts that still get traffic but have aged.
This post was published on {{published:March 2024}}. Review it and list:
facts, versions, screenshots or links that are likely out of date;
sections that no longer match what readers search for; and missing
topics a reader would expect today. Don't rewrite yet. Mark anything you
can't confirm as "check".
{{post}}
Will Google rank a blog post written with AI?
Google's guidance on creating helpful, reliable, people-first content asks whether a page provides "original information, reporting, research, or analysis" and whether it is "written or reviewed by an expert or enthusiast who demonstrably knows the topic well". It also describes using "generative AI to produce large amounts of text without manual oversight or curation" as little to no effort.
That points to how to use these prompts: the model helps you structure, draft and edit, and you supply the experience, the examples and the final review. The SEO prompts cover keyword research, briefs, titles and meta descriptions once the post is written.
How do you keep AI blog posts from sounding generic?
Give it your voice. Paste a few paragraphs you wrote and ask the model to describe your style before it drafts. Anthropic's prompting best practices call examples "one of the most reliable ways to steer Claude's output format, tone, and structure".
Give it specifics. Real numbers, real mistakes and real code from your work are the parts no other post has. Placeholders like [NEED EXAMPLE] in the draft are a feature: they show where you still owe the reader something.
Ban your clichés. If you never want "dive into", "game-changer" or "in today's fast-paced world", list them in the prompt.
Make your favorite one a skill
If you outline or edit every post with the same prompt, turn it into a skill with your style rules built in, so Claude, Claude Code or Codex applies them every time. Paste the prompt into the free SKILL.md generator and it becomes a valid SKILL.md. For general writing help, see the writing prompts for Claude.
To keep a blog prompt library you can reach from anywhere, save these in Kipt with their {{blanks}}, tag them by stage (ideas, outline, edit), and press ⌘K to find one. Your library lives as Markdown in a private GitHub repo you own.