Your best posthas 3 more posts in it

By the end of this you'll have your best posts from the last 6 months, ranked, and 3 new posts written from each one.

Most people post their best idea once and go looking for the next one. But a post that beat your normal by 10x is your audience telling you what they want more of, and most of them never saw it the first time.

The skill finds those posts by scoring each one against your own median, so it ranks what did well for you, not what got the most likes. Then it rebuilds each one with a new first line and a new format, using only the facts that were already in it.

What's inside

1. My own run, with my best post turned into 3 new ones
2. The skill and the prompt that runs it
3. Two ways to pull your posts, a scraper that costs under $1 or Claude's own browser for free
4. The 6 ways it rebuilds a post, from one number to a story
5. The posting plan, so nobody feels a rerun

I ran it on my last 40 original posts, from August 16 to October 29, 2026. My normal score is 32, so a post at 10x scored 320 or more.

I took the first one, my study of 2,204 opening lines, and had the skill write 3 new posts from it. The original was a lead magnet, so I asked for standalone posts with no offer. Every number below is from the original post.

42%.

That's how often a post that opened with a joke made it into its creator's top 5%.

We pulled the opening line of 2,204 posts from 53 creators, all of them already in their creator's top 20%, and ranked every type of opener.

Jokes had the best rate of any opener used 40 or more times.

The average across all of them was 28%.

Try it on the post you're writing this week, and keep everything after the first line exactly as it is.

2 out of 51.

That's how many posts that opened on what "most people get wrong" made it into their creator's top 5%.

We ranked the opening lines of 2,204 top posts from 53 creators, and that opener came close to last.

Here's what else lost:

• picking a fight: 19%, against a 28% average
• opening on your own result: 26%

And what won:

• a joke: 42%
• a guy you met: 39%

Every one of these posts was already in its creator's top 20%, so even the losers were good posts with the wrong first line.

Open your next post on someone else.

We pulled the opening line of 2,204 posts that were already in their creator's top 20%, then counted how often each one reached the top 5%.

Posts that opened on a guy the creator met got there 39% of the time.

Posts that opened on the creator's own result got there 26% of the time.

A reader can picture the person you met much more easily than your numbers.

So before you post the win, find the person in it, the client who was stuck or the founder who asked you the question, and start with them.

The skill reads your posts, scores them, shows you the top 10% and asks which ones to recycle.

1. Download recycle-best-posts.zip below and don't unzip it.
2. In Claude, go to Customize > Skills, click +, then Upload a skill, and pick the zip.
3. Turn on code execution in Settings > Capabilities so it can do the maths.
4. To use Claude's browser, use the Claude desktop app and sign in to LinkedIn in Claude's browser.

Use the recycle-best-posts skill on my profile:
{your LinkedIn profile link}

Read my original posts from the last 6 months in your browser, go slowly, then show me my top 10% before you write anything.

The faster version

Pull your posts with harvestapi/linkedin-profile-posts on apify.com and attach the export instead. It charges about $2 per 1,000 posts, so 6 months of yours costs well under $1.

{
  "targetUrls": ["https://www.linkedin.com/in/your-name/"],
  "maxPosts": 200,
  "postedLimit": "6months",
  "includeReposts": false,
  "includeQuotePosts": false
}

The skill picks 3 different ones for each post, never the same one twice.

Every version gets a new first line and a different format from the original, written in your voice from your own top posts. It never adds a number, client or result that wasn't in the original. If a version needs a fact that isn't there, it asks you or leaves that version out.

Three versions of the same post in one week reads like a rerun. Spread out, each one reads as its own post.