Most people reach for ChatGPT when they want AI to write social posts. That is fine, but Claude prompts are the ones we keep coming back to for anything that needs a real voice, a large paste-in, or careful rewriting. Claude, built by Anthropic, is quietly one of the strongest tools for social media writing in 2026, and the reason is not hype. It is the way it handles instructions and long context.
This is a practical guide. You will find copy-paste Claude prompts for captions, hooks, threads, and repurposing, plus honest notes on where Claude helps and where it does not. Every example is written to be pasted straight into a Claude chat, edited with your own details, and run.
We are not going to pretend AI writes your whole strategy for you. It does not. But used well, Claude for social media takes the slow, blank-page part of content work and turns it into an editing job, which is faster and usually better.
What Makes Claude Strong for Social Writing
Before the prompts, it helps to know why Claude behaves differently from other models, because it changes how you should prompt it.
- Large context window. Claude models in 2026 handle very large inputs, up to around a million tokens on the Opus and Sonnet tiers. In plain terms, you can paste an entire blog post, a full webinar transcript, or a month of past captions and Claude will actually read all of it before writing. You do not have to summarise first.
- It follows detailed instructions closely. Claude is trained to follow the system prompt and your instructions literally. If you give it constraints (character counts, banned words, a specific structure), it tends to respect them instead of drifting. This is the single biggest reason our outputs need less cleanup.
- It matches voice and tone well. Give Claude three or four examples of how you write, and it does a good job of staying in that register. It is less likely to slip into generic marketing language than some alternatives.
- It is good at nuanced rewriting. Turning a formal paragraph into a casual hook, or tightening a rambling caption without losing the point, is where Claude is genuinely useful. It preserves meaning while changing style.
- It handles big paste-ins without falling apart. Because it reads the whole input, repurposing long-form content into many short posts is one of its best use cases.
Keep those traits in mind. The prompts below are written to lean on them.
How to Prompt Claude Well
The quality of your Claude prompts depends far more on the input than on any secret phrasing. Four habits make the biggest difference.
Give it a style guide and examples. Claude matches what it sees. Paste two or three of your best-performing posts and tell it to study the voice before writing anything.
Set clear constraints. Character limits, number of options, required structure, words to avoid. Claude respects these, so use them.
Ask for options, not one answer. Requesting five variations gives you something to pick from and edit, which beats accepting a single output.
Iterate. Your first prompt is a draft of a prompt. Reply with "tighten these, cut the emoji, make the hooks punchier" and the second round is usually the keeper.
Here is a reusable setup prompt you can paste at the start of almost any session:
You are my social media copywriter. Before we start, study my brand voice from these examples and do not write anything yet.
Brand: [your brand]
Audience: [who you are talking to]
Voice notes: [e.g. direct, a little dry, no hype, no exclamation marks]
Examples of my writing:
1. [paste a post]
2. [paste a post]
3. [paste a post]
Reply only with a short summary of the voice you detected, so I can confirm it before we write.
Confirm the voice summary, correct anything that is off, and then run the writing prompts below in the same chat so Claude keeps that context.
Claude Prompts to Turn a Blog Post Into a Thread
This is where the large context window earns its keep. You paste the whole article and Claude builds the thread from the real content, not a guess.
Here is a full blog post. Read all of it, then turn it into a Twitter/X thread of 8 to 10 posts.
Rules:
- First post is a hook that promises a specific payoff. No "a thread:" label.
- Each following post delivers one concrete idea from the article.
- Keep every post under 280 characters.
- Plain language, no emoji, no hashtags.
- Final post has a soft call to action to read the full piece.
Give me the thread, numbered, then a one-line note on which post is weakest so I can improve it.
Blog post:
[paste the entire article here]
The closing request, asking Claude to name the weakest post, is a small trick that gets you a built-in edit pass. You can do the same for LinkedIn by swapping the character limit for a 1,300-character target and asking for line breaks between short paragraphs. For an Instagram carousel, ask for one idea per slide plus a caption to sit under the whole set.
The reason this works so well on Claude specifically is the full read. A shorter-context tool often skims a long article and writes the thread from the headline and the first few paragraphs, which is why those threads feel thin. Claude pulls points from the middle and end of the piece too, so the thread covers the article rather than just its opening. If a post in the output feels generic, reply with "post 4 is vague, rewrite it around the specific example in the article" and it will go back to the source.
Claude Prompts to Repurpose a Podcast or Webinar Transcript
Transcripts are long, messy, and full of gold. Claude reads the whole thing and pulls the usable moments out.
Below is the full transcript of a [podcast episode / webinar] about [topic]. Read all of it.
Produce 10 social posts I can schedule this week:
- 3 short quote-style posts pulling the sharpest lines (lightly cleaned up, meaning unchanged).
- 3 "one lesson" posts, each teaching a single takeaway in 3 to 5 sentences.
- 2 question posts that spark replies, tied to points raised.
- 2 posts framed as "a mistake people make" based on the discussion.
For each post, label the platform it suits best (X, LinkedIn, or Instagram caption).
Keep my voice: [describe it]. No hashtags. Flag anything you were unsure was accurate.
Transcript:
[paste the full transcript here]
Asking Claude to flag anything it was unsure about is important. A transcript can be misheard by the transcription tool, and Claude will not know a misquote from a real one. That flag tells you exactly what to check against the recording.
Claude Prompts to Rewrite Copy in a Specific Brand Voice
Rewriting is one of Claude's strongest tasks. The key is showing it the target voice with examples, not just describing it.
Rewrite the copy below to match my brand voice. Do not change the facts or the offer.
My voice, shown by example:
- [paste 2 or 3 posts that sound right]
What to change:
- Remove hype words and exclamation marks.
- Shorten sentences.
- Keep it warm but direct.
Give me 3 rewrites at different lengths: one short (under 50 words), one medium, one long. After each, note in one line what you changed.
Copy to rewrite:
[paste the copy]
If you run social for clients or manage several brands, this prompt plus a saved set of voice examples per brand is a small system that keeps every account sounding like itself. If you want to lighten the workload further, some of SocialBooster's free tools handle the mechanical jobs (hashtag sets, username checks, engagement math) so you can spend your Claude time on the words that matter.
Claude Prompts to Plan a Week From Content Pillars
Content pillars keep an account coherent. Claude can turn them into a full week without repeating itself, because it can hold the whole plan in view while it writes.
My content pillars are:
1. [pillar]
2. [pillar]
3. [pillar]
Build a 7-day posting plan for [platform] using these pillars. For each day give me:
- The pillar it belongs to
- The format (single post, carousel, short video idea, or story)
- A ready-to-post caption in my voice
- One line on the intended reaction (save, share, reply, or click)
Spread the pillars evenly. Do not repeat hooks or angles across the week. Voice: [describe].
You will get a week you can actually schedule. Read it once, cut the two weakest days, and replace them with a "reply to my prompt" round rather than accepting all seven as-is. Because Claude is holding the whole week in context, you can also ask it to check its own work: "look back over the seven days and tell me if any two posts are too similar, then fix them." That kind of self-review is hard to do by hand across a full plan, and it is where a large context window quietly pays off.
Claude Prompts to Generate Hook Variations
The first line decides whether anything else gets read. Claude is good at producing many angles on the same idea so you can test them.
I am writing a post about [topic / the specific point]. Give me 15 opening hooks.
Vary the approach across the set: curiosity, a bold claim, a specific number, a common mistake, a short question, and a "here is what nobody tells you" angle. At least a few should work with no emoji and no hashtag.
Keep each hook under 15 words. After the list, pick the 3 you think are strongest and say why in one line each.
Take the three Claude picks as a starting point, but trust your own read too. It does not know your audience the way you do. Save the hooks that perform and feed them back as examples next time, which sharpens future results.
Claude Prompts to Write an Engagement Reply Script
Replying well is underrated growth work. You can prompt Claude to draft a batch of genuine, on-brand replies you customise, rather than leaving generic "great post" comments.
I want to engage thoughtfully with posts in my niche: [niche]. Write me 10 reusable reply templates I can adapt.
Each reply should:
- Add something (a follow-up question, a related point, or a respectful counterpoint).
- Sound like a real person, not a bot. No "great post" or "so true".
- Be short, 1 to 2 sentences.
- Leave a clear blank where I fill in the specific detail from the post I am replying to.
Voice: [describe]. Avoid flattery. Aim to start a short conversation.
Use these as scaffolding, never as copy-paste spam. The blank you fill in with a real detail from the post is what makes the reply land. Templates get you started; your specifics make it honest.
Claude Compared to ChatGPT and Gemini
You do not have to pick one tool forever, but it helps to know where each fits.
- Claude is our default for voice matching, long paste-ins, and careful rewriting. When the input is big or the tone matters, it needs the least editing.
- ChatGPT is fast, familiar, and strong at brainstorming volume. If you want 50 rough ideas to sift, it is a fine starting point, and its image tools are convenient.
- Gemini integrates tightly with Google tools and search, which is handy when you want current references pulled in while you draft.
In practice, a lot of people brainstorm in one tool and finish in Claude. There is no rule against mixing them, and the best workflow is usually the one that gets you to a good post with the fewest passes.
Honest Limits of Using Claude for Social Media
Claude is a strong writing partner, but it is not a strategist, and treating it like one leads to bland, off-target content.
- It still needs your judgement. Claude will happily write a confident post about something. Whether that something is worth posting, timely, or right for your audience is your call.
- It does not know your facts. Numbers, product details, dates, and quotes must be checked by you. Claude can and will get specifics wrong, especially from messy transcripts.
- It is not a strategist. It can execute a content plan beautifully once you have decided the plan. Deciding what to say, to whom, and why is the human part, and it is the part that actually drives growth.
- Voice matching is good, not perfect. Read every output before it goes live. The goal is AI as a fast first draft, not an autopilot.
Used with that in mind, Claude for social media saves real time without flattening your brand. The prompts above are a starting kit. Adapt them, save the ones that work, and keep your own examples fresh so Claude keeps sounding like you.
The Bottom Line
Claude prompts are worth adding to your workflow because Claude does the parts of social writing that other tools fumble: reading long inputs in full, following your constraints, and rewriting in your actual voice. Paste your best posts as examples, set clear rules, ask for options, and iterate. Let Claude turn blank pages into edits.
What it will not do is decide your strategy or verify your facts. Keep those jobs, hand the drafting to Claude, and check everything before it ships. Do that, and AI becomes a genuine accelerator instead of a source of generic filler. Start with one prompt from this guide today, run it on real content, and keep the version that sounds like you.