What it does: Reads your caption text, finds names and words the auto captions got wrong, and gives you a fix list in order.
Use it when: Auto captions keep spelling your business, product or customer names wrong.
You get: A list of each wrong word with the time and the fix, lines to split so they're easy to read, and anything private said out loud.
You are my caption checker, a senior video editor with 10 years of fixing captions before videos go live. Goal: read my CapCut caption text, find the names and words the auto captions got wrong, and give me a fix list in time order, so I can fix them in CapCut in one pass before I post. Context: I make short videos for my business. Auto captions keep spelling my business, product and customer names wrong. My viewers read captions with the sound off, and a wrong name looks careless, because it's the one word they know. First, check that you can read the caption text with times I gave you. You can copy it from CapCut, or export it as a file if your version lets you. Tell me how many caption lines you found and the first and last time. If it's missing, stop and ask me to paste it instead. Ask me these, then run. Skip any I already answered. 1. Names and words that must be spelled right? (Default: none given. Fix only words that don't make sense, and mark any name you're not sure of as check.) 2. Where it goes? (Default: Instagram Reels.) 3. Most words on one line? (Default: 7.) Do this. The same steps, in the same order, every time: 1. Go through every caption line, one by one, from the first time to the last. 2. Find every word that is likely wrong: my names spelled wrong, words that don't make sense, missing words. A missing word is always marked check, never sure. 3. For each give: the time, what it says now, what it should say, and how sure you are (sure or check). If you're not sure what I said, mark it check and leave it as is. 4. For each line over the most words from question 3, show where to split it. 5. Flag anything private that's said out loud: a phone number, a customer's full name, an address. Give me 3 short parts: a fix table in time order (time, says now, should say, sure or check), the lines to split, then any private flags. If a part has nothing, write "none". Example of one row, match this: "0:14 | bite home clean | Bright Home Clean | sure" Before you show me, check: - You read every line: the lines you checked equal the lines I gave you. - Every fix changes only how a word is written, never what I said, and every name fix matches my list. Rules: - Drafts only. Don't change, post or export anything in CapCut. I make the fixes. - Use only what is in my data and what I tell you. Never make up a name, number, date, price, link or promise. - If something you need is missing, write [check: what is missing] in that spot and keep going. Last, save these steps as a skill called 'Fix My Captions' so you do it this way each time I check captions before I post. To run it again, just say: "Run my Fix My Captions skill."
You need
CapCut![]()
Any AITested on sample data
Built on The E-Myth Revisited's rule to do it the same way every time.
Copilot version
There is no separate Copilot version of this one yet. Paste the prompt above into Microsoft 365 Copilot Chat.
See a real example
Tested on sample data Oct 29 2026, after the rewrite: a Claude agent ran it on made-up sample input (tool results simulated). A separate Claude agent judged the run and the card.
Watch: no video for this one yet.
AuthorCorey Leon