It said
"I'll squash-merge the PR once CI is green, then verify the deploy."
It meant
Once the automatic checks pass, I'll make the change official — then double-check it's actually live.
AI coding agents just handed anyone the power to build a real product — no CS degree required. Talking to your agent already works fine. The problem is the direction coming back: status updates, decisions, warnings, all in engineering-speak, because that's the language it learned from.
That matters more than it sounds. You're not a student of this work — you're the manager of it. When the report comes back in a language you don't read, you lose the ability to review the work. You can only approve it.
Dejargonizer makes your agent talk back in plain language by default. Read the report, and you can manage the agent. Manage the agent, and you get a better product.
Not sure which is you? Pick the one that sounds like how you build — same underlying file either way, dejargonizer/SKILL.md.
"Turn it on" just means getting that file in front of your agent, however your tool allows it.
Start a brand-new chat or session — old ones often keep ignoring new rules.
3Ask it, word for word:
4Check what comes back:
Heard this week
Talking to your agent was never the problem. This is about the direction back — the difference between nodding along and actually running the project.
"Hardcoded the pricing constants for now, added a validation lib."
"Two decisions I made for you. First, the prices are typed in by hand rather than read from your spreadsheet — that got it working today, but the page will be wrong the first time you change a price. Second, I brought in a small outside tool to check the form entries instead of writing that part myself. Happy with both, or want either done differently?"
The first kind of sentence saves you a confusing minute. The second saves you a bug in production — because you got the chance to say no before it shipped.