Hey, it's Angelica.
← all resources

The EncoreTurn the post that worked into twenty more

threads skill

You post something. It goes further than usual. You feel good for an afternoon, and then you open a blank box and start again from nothing. That is the part that makes no sense. The post that worked is the only piece of real information you got all month, and almost everyone throws it away the next morning.

What this is. A skill you install in Claude. You paste in one post that performed, answer two questions about it, and it hands back twenty ways to run it again — different shape, different angle, different job, different person it is aimed at. One post in, twenty out.
the whole idea

A post that worked is not a sentence. It is a combination

When something outperforms, the instinct is to think the words were good. Sometimes they were. More often what worked was everything underneath them: a specific tension, pointed at a specific person, in a shape that made it easy to agree with out loud.

That combination is the asset. The words are just one arrangement of it — and you tested exactly one. There were a dozen other arrangements of the same true thing, and you have no idea which of them would have gone further, because you never wrote them.

So the job is not “post it again”. The job is: hold the part that was proven, change one thing, and send it back out.

Do that twenty times and you are not repeating yourself. You are running twenty experiments on a hypothesis that already has evidence behind it, while everyone else is starting from zero every single morning.

what it changes

Six things it varies, one at a time

Every idea it gives you back keeps the proven part and moves exactly one variable. That is deliberate: change three things at once and when the new one lands you have learned nothing about why.

The shapeSame claim, built as a Normalize, a Callout, a Flat Truth, a Hill. The structure does a surprising amount of the work, and your original only used one of them.
The angleZoom in on the one detail everyone replied to. Zoom out to the pattern behind it. Flip it and argue the other side. Say what it costs you to believe it.
The jobYour original had one: reach, or trust, or replies, or a sale. The same material can be pointed at any of the other three, and it reads completely differently when it is.
The evidenceSwap the proof. A story becomes a number, a number becomes a screenshot, an opinion becomes the specific moment you changed your mind about it.
Who it is aimed atThe same truth, written for the person one ring further out. Or one ring further in, where it gets sharper and fewer people recognise themselves — which is usually the version that travels.
The formatOne line. A list. A question with a real answer. A before and after. The same thing at four words and at ninety.
why this is not lazy

Repeating a winner is the most reliable thing on the platform

I went looking for this in the data rather than assuming it. Across 20,739 text posts from 33 creators, twenty of the twenty-five most consistent accounts republish the same text, word for word, weeks apart. There were 1,131 repeated texts in the set, and in 304 cases two or more copies each went viral independently.

One creator posted the same line about job interviews twenty-one times. Twenty of them hit. Nobody complained, nobody noticed, and the only person who thought it was repetitive was the person writing it.

A post that beat your median once has already proved there is an audience for that idea. Variations of it are the highest-odds content you can make.
A brand-new idea every day is not discipline. It is throwing away your only evidence and calling it originality.

Straight reposting works. This takes it further: instead of the same post twice, you get twenty near-relatives of it, which means you can run the winner for a month without anyone seeing the same sentence twice.

before you paste

Which post to feed it

Not your all-time best. The one that beat your own normal by a clear margin and that you could happily say again.

  1. Three times your median or better. Your median, not anyone else's. A post at triple your usual has told you something; a post at 1.2 times has told you nothing except that it was a Tuesday.
  2. At least a week old. Numbers keep moving for days. Judge too early and you will pick the wrong winner.
  3. Text, ideally. If the performance came from an image, what you proved is that the image worked, and that does not remix.
  4. Something you actually believe. You are about to say it twenty more ways. Pick the one you would defend out loud, not the one that happened to catch.
No idea which of your posts did best? That is a solved problem: Viral Post Research shows you how to sort any Threads profile by performance in about thirty seconds.
take it

Download

the-encore.skill Claude skill · [[serve: il file .skill e la sua dimensione]]
This one needs Claude. It is a skill, which is a thing you add to Claude so it knows how to run a specific job properly. It will not work in ChatGPT — the file is instructions written for Claude to follow. The free plan is enough.
five minutes, once

How to install it

All in the browser. Nothing to install on your computer, no terminal, no code.

  1. Download the file above. It lands in your Downloads as the-encore.skill. Do not try to open it — it is not meant to be read, it is meant to be uploaded.
  2. Go to claude.ai and sign in. A free account works.
  3. Open Settings, then Capabilities, then Skills. The wording moves around a little as they update things — you are looking for the Skills section in your settings.
  4. Upload the file. It appears in your list of skills and stays there. You do this once, ever.
  5. Open a new chat and paste your post. “This one did well, give me the encore.” That is enough.
If it doesn't start on its own, name it: “use the encore”. Works every time.
what comes back

Twenty ideas, ranked, with the reasoning attached

Not twenty paraphrases. Each one tells you which variable it moved and why that move is worth testing on this particular post, so you can throw out the eight that are wrong for you and still know what you are doing with the other twelve.

  1. It reads the original first. Before it writes anything it tells you what it thinks actually worked — the tension, the shape, the person it was aimed at. Correct it if it is wrong. Everything downstream is built on that one paragraph.
  2. Twenty ideas, in order of odds. Ranked by how close they stay to the proven part. The top of the list is the safe money; the bottom is where the interesting failures live.
  3. Written, not described. Each one comes as a post you could publish, not a brief telling you to write a post.
  4. A running order. How to space them so the same idea does not show up twice in one week.
so you are not disappointed

What it does not do

It does not invent a voice. If you hand it a post that does not sound like you, it will cheerfully produce twenty more that do not sound like you either.

It also will not tell you a post was good when it was not. Paste in something that did average numbers and it will say so and ask for a better one, which is annoying exactly once.

More free resources Work with me