Review code directly from your terminal
npm install nero-review
src="https://raw.githubusercontent.com/alcanivorax/nero-review/main/assets/nero-review-preview.gif"
alt="nero-review demo"
width="800"
/>
nero-review is a terminal-first AI code review tool.
It analyzes source files, understands their role and structure, validates AI output,
and delivers clear, structured, actionable feedback ā without pulling you into a browser or UI.
Built for developers who live in the terminal.
Install globally using npm:
``bash`
npm install -g nero-review
nero-review reviews one file at a time.
Run it from your project root and pass the path to a single source file:
`bash`
nero-review
`bash`
nero-review ./api/login/route.ts
nero-reivew currently supports OpenRouter as its AI provider
You must configure and OpenRouter API key and model before running the CLI.
---
Create a secrets file:
`bash`
~/.config/nero/secrets.sh
Add your configuration:
`bash`
export NERO_API_KEY="your_api_key_here"
export NERO_MODEL="your_model_name"
Then source it in your shell config (.zshrc, .bashrc, etc.):
`bash`
[ -f "$HOME/.config/nero/secrets.sh" ] && source "$HOME/.config/nero/secrets.sh"
Set environment variables using PowerShell:
`bash`
setx NERO_API_KEY "your_api_key_here"
setx NERO_MODEL "your_model_name"
Restart your terminal or VS Code after setting them.
nero-review performance depends heavily on the selected model.
For the best experince, use one of the following OpenRouter models:
---
Best choice for everyday use. Good speed with reliable reasoning.
Free
`text`
mistralai/devstral-2512:free
google/gemma-3-27b-it:free
Paid
`text`
anthropic/claude-3.5-sonnet
openai/gpt-4o-mini
Free
`text`
qwen/qwen3-coder:free
Paid
`text`
anthropic/claude-3-opus
openai/gpt-4o
`bash``
File ā Analyze ā Prompt ā AI ā Validate ā Format ā Output
- API keys are read only from environment variables
- No keys are logged or printed
- No data is stored or cached
- Requests are sent only for the file under review
Contributions are welcome! š
If you'd like to improve the project (documentation, fixes, or small enhancements),
please follow the steps outlined in CONTRIBUTING.md.
This project is beginner-friendly, and small improvements are always appreciated.