Stop Re-prompting.
Start Kolewoofing.
Koleslaw takes your rough prompt and turns it into exactly what AI needs to hear. Better input, better output — it's not rocket science. It's a barking cow.
Enhance in place on ChatGPT, Claude, and Gemini.
How it works
Paste a draft and get back a prompt an AI model can act on. Koleslaw reads the shape of what you wrote and picks one of three strategies.
Expand
A short, terse prompt gets the specifics a model would otherwise guess at: who it's for, the tone, how much detail, the output format, and the limits.
Refine
A prompt that already says enough gets tightened. Vague words become precise ones, filler goes, and the instructions that matter most move to the front. Your voice stays.
Structure
A long, rambling prompt gets reorganized so it can be skimmed: the objective first, then context, requirements, and output format. Nothing you asked for is dropped.
Where it works
One API behind every client. Use whichever is closest to where you write prompts.
The playground on this page
Free, no account, 10 enhances a day. Paste, enhance, copy.
Chrome extension
On ChatGPT, Claude, and Gemini, a floating button rewrites the draft in the prompt box in place. The popup handles every other site: paste, enhance, copy. Get it on the Chrome Web Store
Claude Code
A plugin that adds a /koleslaw:enhance command and a skill Claude Code picks up from plain language, so a half-formed request gets rewritten before it runs. Plugin on GitHub
The API
Streaming and synchronous endpoints behind an API key, with the strategy on auto or set explicitly per request. The server is open source. API on GitHub
What runs it
Our own fine-tuned model
Koleslaw is served by a 31.6B-parameter fine-tune of NVIDIA's Nemotron, trained for prompt enhancement and run on a single GPU in AWS. The weights are public. Open weights on Hugging Face
Built in the open
Notes from running a fine-tuned 30B model in production: quantization, spot GPUs, failure design, and what it all costs. Read the Koleslaw Blog
Questions
Is Koleslaw free?
The playground is free with no account: 10 enhances a day. A free account gets 50 a day per API key. Pro is $10 a month for 500. Pricing
Will it change what I'm asking for?
No. Every strategy is instructed to preserve your intent and your tone. It adds what a model would otherwise have to guess, or tightens and reorganizes what you wrote.
Do I have to pick a strategy?
No. Koleslaw picks expand, refine, or structure from the length and shape of your draft. The API lets you name one explicitly.
What happens to my prompt?
It goes to the Koleslaw API, gets enhanced, and comes back. It may be cached for up to 24 hours so a repeat comes back fast, then it is purged. It is never used to train a model. Privacy policy