Can You Spot Your Own Voice? Run a Free AI Slop Test on Threads

You open Threads for five minutes and the posts start to blur: a "hot take" that isn't hot, a tidy three-line lesson, "here's the thing", a closing question nobody asked for. Ten posts in, it all sounds like one polite, slightly beige voice that could have come from any chatbot.

Then you open the post you scheduled for tomorrow and get a small sinking feeling. Would anyone know you wrote this? Would you, if your name wasn't on it?

That doubt changes how you write. You sand off the rough lines that sounded most like you because they felt too blunt, or you post the polished version and wonder why the replies have gone quiet. It can happen even if you write every word yourself, because reading so much AI-written content lets its habits creep into your own posts.

This free workflow gives you a quick, honest check. Your AI agent takes five of your real Threads posts from the last 30 days and writes a plain-AI twin of each one on the same topic. It shuffles all ten, and you call each card human or slop without knowing which is which. Then you see a slop score for your posts next to the twins' score, and Threadify rewrites the worst-scoring twin from what it knows about you, so you can see that topic written in your voice. Nothing is scheduled until you say yes.

If you'd rather see whether your raw posts get more views than your polished ones, score your past posts for slop instead. This test is about your voice and whether you can still pick it out.

What you need before you take the slop test

• A Threadify account with your Threads account connected. The test reads your recent posts from there, so you need some posts from the last 30 days.
• An AI agent connected to Threadify. Agent connections come with every paid Threadify plan. The example run used the Claude desktop app.
• A few quiet minutes to read ten short posts and judge each one by feel.

How to test whether your Threads posts sound like AI (6 steps)

Step 1: Let your agent read your Threads posts

This is what lets your agent pull your real posts for the test. In Threadify, open the Agent tab at the top, scroll to Set up a client and pick your agent. For Claude, click Copy next to the Connector URL. Then in Claude, open Settings → Connectors → Add custom connector, name it Threadify, paste the link and leave Client ID and Secret blank. Claude opens your browser, where you log in and click Allow. If you get stuck, the full Claude connection guide walks through it.

The Agent tab with the Claude card. Copy the connector link, then follow the three numbered steps underneath.

Step 2: Install the free slop test with one message

One message installs Threadify's free workflows and asks your agent to run the slop test, so there's nothing to set up by hand. Paste this into your agent:

install this and run threadify slop score https://github.com/lennoxsaint/Threadify-workflows

If you've already installed the workflows for something else, such as finding your niche from your own posts, skip to Step 3.

An install message for a different workflow, to show the format: the workflows link and one plain sentence saying what you want. Yours asks for the slop score.

Step 3: Start the blind test with one command

The command sets up the test and keeps the answers hidden from you. When your agent says it's ready, send the command shown typed in the message box in the screenshot, /threadify-slop-score.

Your agent then picks five of your recent posts and writes a plain-AI twin for each one. A twin is what an AI writes when you just ask it to "write a Threads post about [topic]" and tell it nothing about you. The twins are only for the test and are never posted.

The agent has read its briefing and is waiting for the command, which is already typed in the message box. The example run used Opus 5.5 at medium reasoning.

Step 4: Call each shuffled card human or slop

This is where you find out whether you can still recognise yourself. Each card shows a post with no author label, numbered like "Card 2 of 10". Read it and decide whether you wrote it or a plain AI did.

Go with your gut. The test scores four tells: hedging (softening a point instead of saying it with confidence), stock AI phrases, a corporate voice and over-formatting.

In the example run, one twin opened by explaining that reading "isn't just a hobby". The real post on the same topic simply called books a cheat code for life.

Card 2 of 10 from the example run, with no name attached. This one was a real post: short, lowercase and blunt.

Step 5: See your slop score next to the AI twins

After the last card, the agent reveals the answers and scores both sets of posts. Each hedge costs 30 points, every other tell costs 10, and no post scores above 100. One hedge on its own puts a post at 30, so if your own posts score close to the twins, or you mixed them up, hedges are the first thing to cut.

In the example run, all ten calls were right, with the real posts "averaging a slop score of 0 versus 16 for the plain AI twins". The worst-scoring post was the twin about learning to code, at 30. The checklist under the result shows what comes next: rewrite the worst twin using what Threadify knows about you (your Brain), add a follow-up reply called an Auto Plug, then show you both and wait for your yes.

The reveal from the example run: 10/10 right, 0 versus 16, and the worst post at 30. The checklist shows the fix step is next.

Step 6: Review the rewrite of the worst post before anything is scheduled

Threadify rewrites the worst-scoring twin from your Brain (what Threadify knows about you), so you can see the same topic written the way you'd write it. The agent shows you the rewrite and waits. Nothing goes on your schedule until you say yes.

In the example run, the rewrite opened with a specific, personal story about learning to code for free and rewriting a site 47 times. The approval message showed check PASS, and the rewrite was reported as scoring 0, down from 30. The hook was good, but the last two sentences fell flat. Read the rewrite out loud, and if any line doesn't sound like you, don't say yes yet.

The approval message may also include an Auto Plug: a follow-up reply set to post 15 minutes after the main post. In the example, the plug promotes the example account's own product, so check that yours points to something you actually offer before you approve.

The approval message from the example run: the rewritten post, the Auto Plug set for 15 minutes later and the check PASS label. Nothing is scheduled yet.

What to expect from your first slop test

• It's a quick self-check, not proof of anything. Treat a perfect score as a good sign and nothing more.
• Your "real" posts may not be fully human. In the example run, one post called human had been drafted with help from a Grok bot. If you've used AI on past posts, some of that may show up on your side of the score, though in the example run it didn't move the average.
• The example run used the Claude desktop app with Opus 5.5 at medium reasoning. The install instructions also name Claude Code and Codex, but the example run didn't try other agents.
• The rewrite can still contain slop. Threadify is tweaking how it writes drafts so they sound less like AI, so read every rewrite yourself before it goes anywhere. If rewrites keep missing your voice, give Threadify more of your own stories and phrases to work from.
• It checks how your posts sound. It doesn't look at views or replies; the slop check linked above compares raw and polished posts on views.

Find out if your posts still sound like you this week

Connecting your agent is what lets it read your posts and hand the rewrite to Threadify. Open Threadify and connect your agent from the Agent tab, then paste the install message, take the blind test and decide for yourself whether the rewrite sounds like you.