Rewrite the prose that smells generated or padded.
Smell Check finds the habits that make writing sound generated or padded: the "it's not just X" pivot, the hope-this-finds-you-well opener, the claim nobody could check. Then it rewrites them and keeps your voice.
It is an installable writing skill for AI agents. You get revised text back, not a report.
Before / after
A job announcement
Before. I'm thrilled to share that after an incredible journey, I've officially taken the next step in my career and joined Acme as Head of Partnerships. This isn't just a new role — it's a chance to shape the future of how teams work together.
After. Some news: I started at Acme this week as Head of Partnerships, and I'm thrilled about it.
The feeling stays. "Incredible journey," "next step," and "isn't just a role, it's a chance" are the parts that read as generated.
An email that buries the ask
Before. I hope this message finds you well. I wanted to reach out to let you know that, unfortunately, we will need to reschedule Thursday's meeting. I truly apologize for any inconvenience this may cause, and I appreciate your understanding and flexibility.
After. I need to move Thursday's meeting. Sorry for the shuffle. What time works for you instead?
The ask was buried in the middle. Now it comes first, and the apology is one human sentence instead of three formal ones.
A cover-letter claim
Before. I have extensive experience leading high-impact teams.
Smell Check asks. How many people, for how long, and what changed?
Your answer. I ran a crew of four at the county pool for two summers. We cut the swim-lesson waitlist in half.
After. I ran a four-person crew at the county pool for two summers and cut the swim-lesson waitlist in half.
Smell Check does not invent the numbers. When a claim has nothing behind it, it asks.
A strong word that stays
Kept. The renovation transformed the kitchen: we took out the wall, and now it's one room.
A strong word stays when it points at something you can see. That is the earn-the-word test, not a banned-word list.
Install in your agent to try it on your own draft.
What it reads, writes, and changes
| Reads | A page, essay, note, or other publishable prose |
| Writes | Edits in the file you named, or the revised prose in chat when you paste it |
| Changes | The prose. Smell Check is a spray, not a report-only pass |
Boundaries
Smell Check does not guess whether a human or a model wrote the sentences, and it does not try to fool detectors. It goes after the style people now call AI writing, whoever typed it. Fuller positioning: About.
Which pass
Use the claims pass when the problem is truth or scope. Use the prose pass when the meaning is sound but the wording gets in the way. Add a house-style overlay when your publication has its own rules.
The skill reads shared law, then a genre file if the surface needs one, then your project overlay if you wrote one. The bundled defaults are enough for a first run. An agent reads the rules and edits; nothing scans the tree, and v1 has no CLI. The file map and load order live in Files.
The npm package also exports readRule so a script can load the same files. That helper does not score prose.
Built by Catalyst Forge LLC. MIT.