Social Repurposer: one idea, drafted for every channel

★

Write the idea once, show up everywhere

You have the idea. You said it on a podcast, wrote it in a blog post, or just talked it through in a voice note on your drive home. Turning that one idea into a LinkedIn post, an Instagram caption, and an X thread means rewriting it three times, each with its own length, tone, and hashtag habits. So most of it never gets posted, and the channels where your customers actually are go quiet.

This kit does the rewriting. You hand it one source (a post, a transcript, or a voice-note text) plus a short brand-voice file you own, and it drafts a tailored version for each platform: the right length, the right tone, sensible hashtags. They are drafts only. You read them, tweak what you want, and post (or schedule) yourself.

Plain-English glossary, once: Your source is whatever you already made: a blog post, a call transcript, or a voice note typed out. Your brand-voice file is a plain file you own and edit: a few lines on how you sound, words you like, words you avoid. Claude is the AI that reads both and writes the per-platform drafts. The brand-voice file is what keeps the drafts sounding like you, not like generic AI. Want sharper inputs? Lesson on prompting shows you how.

1

Drop in your source: the idea you already made

What's happening: The repurposer starts from something real you already created, so the drafts have substance instead of fluff. You paste one source into data/source.md (or point the kit at a file): a blog post, a podcast or call transcript, or a voice note you typed or auto-transcribed. One idea in, three platform drafts out.

What you can feed it	Where it lives	Example
blog post	data/source.md	A 600-word article you published this week.
transcript	data/source.md	The transcript of a podcast or sales call.
voice note	data/source.md	A 90-second idea you dictated, typed out.

Source file goes here (the kit's setup wizard asks for this path, not saved on this page)

↳ How to feed it a clean source →

2

Add your brand voice: the file that makes drafts sound like you

What's happening: This is the part that makes the drafts yours, and it's yours to edit. Your brand-voice file is a short, plain file (data/brand-voice.md) describing how you sound: your tone, words you love, words you avoid, and whether you use emoji. The kit reads it and writes every draft to match. Change one line here and every platform draft shifts with it.

In brand-voice.md	Example line
tone	Warm, direct, a little funny. Talk like a smart friend, not a brochure.
use	Short sentences. Real examples. The word "honestly".
avoid	Hype words (revolutionary, game-changing), jargon, exclamation overload.
emoji	A few on Instagram, almost none on LinkedIn, none on X.

Brand-voice file goes here (in the kit's setup wizard, not saved on this page)

Why this keeps you in control: the brand-voice file lives on your machine and is owned and edited by you. The kit never invents a voice or borrows someone else's: it writes only to the rules in your file. No file, no made-up tone. You can rewrite it anytime and re-draft in seconds.

3

Give it a brain: connect Claude

What's happening: The source gives the substance, the brand-voice file gives the sound, and Claude does the writing. The kit asks Claude for a structured result (a typed object: one draft per platform, with body, hashtags, and any claims it thinks you should fact-check), never loose prose. It is instructed, in plain rules, to rewrite per platform and to flag, not invent, anything it cannot verify from your source.

Go to console.anthropic.com and sign in.
Open API Keys → Create Key.
Copy the key (it starts with sk-ant-). The setup wizard saves it on your machine.

Your Claude API key goes here (in the kit's setup wizard)

Default brain: claude-opus-4-8 (sharpest). High volume? Set claude-haiku-4-5 to cut cost.

4

Try it: draft for three channels, send nothing

What's happening: Paste an idea (or tap a sample). The repurposer pulls the core point, then writes a tailored draft for LinkedIn, Instagram, and X, each with the right length, tone, and hashtags, and flags any claim worth checking. Then it waits. This is confirm-before-post: it drafts, you review. This repurposer is live. It runs entirely in your browser, nothing is sent anywhere.

Social Repurposer · brand voice: warm, direct (sample)

Tap a sample idea, or paste your own:

Paste a post, transcript, or voice note

Pure in-browser dry run: no API call, nothing posted, nothing saved.

5

You review, then you post: it never posts for you

What's happening: The kit stops at drafts. It writes the three versions, flags anything you should double-check, and hands them to you. Nothing is published. You copy a draft, tweak it, and post it yourself, or drop it into whatever scheduler you already use. The drafts above are wired to exactly this: try the review buttons.

Four promises the kit keeps:
• Drafts only. It never auto-posts. Nothing reaches a feed without you posting it.
• The brand-voice file is yours: it writes only to your rules, never a borrowed voice.
• It flags claims that may need fact-checking instead of stating shaky numbers as fact.
• Nothing is sent or scheduled from the kit unless you wire it to your own scheduler on purpose.

Want a second set of eyes on the drafts before you post? Lesson on spotting bad output is a quick checklist for catching off-voice lines, shaky claims, and anything that sounds like a robot wrote it.

6

Run the whole pipeline from one command

What's happening: Day to day you do not touch any of the wiring. You record or write an idea, drop it in data/source.md, run one command, and read three drafts. The kit ships with DRY_RUN on, so it drafts and plans but never posts until you decide otherwise.

npm run dryrun     # draft all three platforms in your terminal, nothing posts
npm run draft      # write drafts to ./out as plain files you can copy
npm run voice      # check a draft against your brand-voice file
npm start          # run it for real (still drafts only, you post)

Drafts only, with a kill switch: the kit never has a "post" button of its own. Even wired to a scheduler later, it produces drafts and queues them for your approval, so you watch it work before anything reaches a real feed.

Run it yourself: it's free and it's the real thing

The kit is a complete, runnable, MIT-licensed repo: one source in, a brand-voice file you own, and tailored LinkedIn, Instagram, and X drafts out, with claims flagged for you to check. Runs on your machine, drafts only, nothing crippled. Run npm run setup and you're repurposing.

Get the free kit on GitHub →

▶

Go live

With the kit downloaded:

npm install
npm run setup     # Claude key + brand-voice file saved on YOUR machine
npm run draft     # turn data/source.md into three platform drafts

Write your brand-voice file once, then for every new idea just update data/source.md and re-run. Prefer the command line first? npm run dryrun drafts to your terminal without touching a single feed. For sharper inputs and outputs, pair it with the prompting lesson and spotting bad output.

Want an always-on content engine wired to your scheduler?

This kit is the DIY taste. We build your always-on content engine: it watches for new sources (your blog, your call transcripts, your voice notes), drafts every platform in your real brand voice, and feeds the queue straight into your scheduler for one-tap approval, so your channels stay alive without you rewriting anything three times.

Have Rabbithole build it → ↳ Pair it with spotting bad output