LinkedIn Repurposing Prompt Chain
A set of independent posts extracted from one source, each standing on its own claim, with an honest count rather than a forced ten. Repurposing usually means slicing one article into ten thin posts. This chain does the opposite: it finds the claims the source can genuinely support, discards the ones it cannot, and writes each survivor as a standalone post. Four steps, run in order, each feeding the next.
Step 1 of 4 — Extraction.
You are an editor reading source material for claims that could each carry a post on their own.
## Source
{{SOURCE_MATERIAL}}
## Context
Audience: {{AUDIENCE}}
The position this should ladder into: {{TOPIC}}
## Task
List every claim the source genuinely supports. For each: the claim in one sentence, the evidence inside the source that supports it, and a strength rating (strong / adequate / thin).
Then state how many posts this source can honestly carry. If the answer is two, say two. Do not stretch it.
## Output
| # | Claim | Evidence from the source | Strength | Stands alone? |
|---|---|---|---|---|
Then: "This source supports N posts" plus one line on what was too thin and why.Step 2 of 4 — Angles.
Take the claims rated strong or adequate from step 1.
For each one:
- The angle: what this post argues, in one sentence, phrased so a reader could disagree.
- Who it is for: the specific reader, not "B2B marketers".
- Why now: what makes it worth saying this month rather than any month. If nothing does, mark it evergreen and say so honestly.
- What it must not claim: the overreach the evidence does not support.
## Output
| # | Angle | Reader | Why now | Must not claim |
|---|---|---|---|---|
Then order them for publishing, and say why that order. Usually: strongest first, because the first post decides who sees the rest.Step 3 of 4 — Drafts.
Write each post from step 2.
Rules:
- 120–220 words.
- Opening line states the claim or the surprise. No teasing, no formula openings.
- One concrete detail from the source, with its real numbers.
- State the limitation of the claim in one line.
- End with something answerable.
- Never add a fact that is not in the source material from step 1.
- No hashtags beyond two, no emoji bullets, no engagement bait.
## Output
For each post: the full text, the word count, and the one line from the source it rests on.Step 4 of 4 — Schedule and critique.
Take the drafts from step 3.
1. Order them across the next N weeks at my stated cadence, with at least four days between posts that make related claims — two posts arguing the same thing in one week reads as a campaign, not a point of view.
2. For each post, name its single weakest sentence and say what is wrong with it.
3. Identify the one post that should not be published and say why.
4. Name what is missing: the post this set should contain but the source could not support, and what material would create it.
## Output
### Schedule
| Date | Post | Claim | Weakest sentence |
|---|---|---|---|
### Do not publish
Which one and why.
### The missing post
What it would argue, and the material needed.- You have a webinar, a long call, a report or an article and nothing on the calendar.
- You want a month of posts from work you already did.
- You are running content for a client and the source material is what they gave you.
- Run the steps in order in the same conversation. Each one reads the previous output.
- Stop at step 1 if the count is low. A source that supports two posts supports two posts.
- Do step 4 even when you are in a hurry. "Which one should not be published" is the highest-value line in the chain.
The output schema it demands
The single highest-leverage part of any prompt. Without a declared shape you get prose you then have to reformat; with one, you get sections you can act on — and you can tell at a glance when the model skipped something.
Step 1 → claims table + honest post count Step 2 → angles table + publishing order Step 3 → full drafts + word counts Step 4 → schedule + weakest sentence per post + the one not to publish
What it looks like filled in
Illustrative, not a promise — this is the shape of a good answer rather than a guarantee of one. The scenario is the same fictional company used across the whole toolkit.
Source: a 40-minute webinar transcript on multi-entity close best practices.
Step 1 output — 7 claims extracted, 2 rated strong, 3 adequate, 2 thin. "This source supports 5 posts" — not the 10 a transcript that length usually gets stretched into.
Step 2 output — the strongest claim goes first, because it decides who sees the rest; the thin two are dropped rather than padded out.
Step 4 output — a schedule across 3 weeks with a 4-day minimum gap between related claims, and one post flagged "do not publish" — the ERP-migration-timing claim was not actually stated clearly enough in the transcript to stand alone as a post.
Attach these once and stop retyping your business
This prompt works pasted into a blank chat. It works considerably better inside a Project with these attached, because then the ICP, the product and the limits are already in context and every answer inherits them.
The reference behind this prompt
If you repurpose other people’s material, take the source properly rather than the summary someone else already wrote.
Written to a fixed structure and run through Claude and ChatGPT to confirm each one returns the output format it declares. That is not the same as a measured reply-rate benchmark, and nothing here claims one — if you run these at volume and have numbers, we want them.
Take the whole folder instead
This prompt, the files it expects and the rest of the sequence, as a zip you can drop into a Claude Project, a Cursor workspace or an agent repo. Free, no account.