A small terminal app that proofreads/rewrites text using any OpenAI-compatible chat-completions endpoint — Ollama, LiteLLM, vLLM, OpenAI itself, etc. Edit in a text area, pick a mode, hit a shortcut, get corrected text back in place with changes highlighted.
- Proofread — fix spelling/grammar, keep tone and length as-is
- Professional — rewrite in a business-appropriate tone
- Casual — relaxed, conversational tone
- Polite — extra courteous tone
- #social — punchy social post with hashtags
Output stays in the same language as the input.
Requires Go 1.26+ and Ollama running locally with a model pulled:
ollama pull gemma4:e4b # or any model you like
ollama serve # if not already runningBuild:
go build -o proofread .Or install straight from the repo (puts proofread in $(go env GOPATH)/bin):
go install github.com/StevenCyb/proofread@latest./proofread # uses gemma4:e4b on localhost:11434 (Ollama)
./proofread -model gemma4:26b # bigger model, better quality, slower
./proofread -host https://my-litellm.example.com -model gpt-4o -api-key sk-...Any backend that speaks the OpenAI /v1/chat/completions API works — point -host at it and pass -api-key (or set PROOFREAD_API_KEY) if it requires auth. Ollama serves this same API locally, so no separate mode is needed.
| Key | Action |
|---|---|
ctrl+n / ctrl+b | next / previous mode |
ctrl+p | proofread current text |
| any key | dismiss highlight, resume editing |
esc / ctrl+c | quit |
While a request is in flight the border pulses and input is blocked. When it finishes, the text area is replaced with the corrected text; changed words are underlined until you start typing again.
Each setting resolves in this order, highest priority last:
- built-in fallback
.envfile next to the binary (not the working directory)- real environment variable
- CLI flag
| Flag | Env var | Fallback | Description |
|---|---|---|---|
-host | PROOFREAD_HOST | http://localhost:11434 | OpenAI-compatible API host (Ollama, LiteLLM, etc.) |
-model | PROOFREAD_MODEL | gemma4:e4b | Model name |
-api-key | PROOFREAD_API_KEY | (empty) | API key, if the endpoint requires one |
Example .env dropped alongside the proofread binary:
PROOFREAD_HOST=https://my-litellm.example.com
PROOFREAD_MODEL=gpt-4o
PROOFREAD_API_KEY=sk-...