Correctly

A minimalist browser extension that checks grammar, spelling, and punctuation using AI — either your own API key or Chrome's built-in Gemini Nano.

Features

• Works on any text input, textarea, or contentEditable element
• Inline correction tooltip with accept/dismiss per suggestion
• Supports OpenAI, OpenAI Compatible (Groq, OpenRouter, etc.), Ollama, LM Studio, and Chrome's built-in Gemini Nano ("Chrome Free AI")
• Currently supports English — more languages coming
• Custom model selection - use any model your provider supports
• Chrome Free AI runs entirely on-device — no API key needed, no data leaves your machine
• Configurable log verbosity for debugging
• Respects spellcheck, disabled, readonly, ARIA attributes, and data-correctly opt-out

Install

1. Download target package from latest release:

   • correctly-chrome.zip for Chrome/Chromium
   • correctly-firefox.xpi for Firefox

2. For Chrome/Chromium:

   • Open chrome://extensions
   • Enable Developer mode
   • Click Load unpacked and select unzipped folder

3. For Firefox :

   • Drop the correctly-firefox.xpi in firefox

Using Providers:

• Ollama: see Using Ollama below
• LM Studio: see Using LM Studio below
• OpenAI Compatible: see Using OpenAI Compatible below

Build locally

• npm run build:release:chrome -> correctly-chrome.zip
• npm run build:release:firefox -> correctly-firefox.xpi
• npm run build:release builds both

Using Ollama

Correctly supports Ollama for local grammar checking. No API key is needed for local instances — if you use Ollama with authentication, enter your key as usual.

1. Pull a model: ollama pull llama3
2. Install and load the extension. The Origin: chrome-extension://... header is handled automatically.
3. If you get a 403 error, your Ollama version's CORS check is blocking the extension origin. Fix:

   # Kill the Ollama app, then:
   OLLAMA_ORIGINS=* ollama serve
   

4. In the extension popup, select Ollama, choose a model, and save. An API key is only needed if your Ollama instance requires authentication.

Using LM Studio

Correctly supports LM Studio for local grammar checking. No API key is needed.

1. Open LM Studio, load a model, and start the local inference server (default port 1234)
2. Make sure Local CORS is turned off in LM Studio's settings
3. In the extension popup, select LM Studio, choose a model, and save

Using OpenAI Compatible

Correctly supports any service that offers an OpenAI-compatible API (e.g., Groq, OpenRouter, DeepSeek). You provide the base URL and API key.

1. In the extension popup, select OpenAI Compatible
2. Enter the full base URL (e.g., https://api.groq.com/openai/v1)
3. Enter your API key
4. Select a model, click Save — the extension verifies the connection and ready to serve

Want to add a new Provider?

• OpenAI-compatible API (e.g., Ollama, LM Studio): extend AbstractOpenAICompatibleProvider — _doCorrectGrammar() and response parsing are already implemented. Just provide static metadata and set this.endpoint in the constructor.
• Generic OpenAI-compatible (any service with a base URL + API key): no new class needed — use GenericOpenAIProvider which is already registered. Users configure the base URL and API key in the popup.
• Other providers: extend AbstractProvider directly and implement _doCorrectGrammar(text) and all required static metadata.

Then add the class to PROVIDER_CLASSES in provider-registry.js.

Privacy and Security

• Your API key is stored locally in browser extension local storage on your device. It is never sent to any server other than your chosen AI provider.
• Chrome Free AI runs entirely on-device — text is processed by Chrome's built-in Gemini Nano model. No data is ever sent over the network.
• For other providers (e.g., OpenAI), text you type is sent to the chosen AI provider for grammar checking. Avoid typing sensitive information in fields where the extension is active, or use data-correctly="false" to opt out specific elements.
• Password fields, credit card inputs, and other sensitive field types are automatically excluded.

Project Structure

correctly/
├── index.html             # Static landing page for the project
├── landing.css            # Landing page styles
├── landing.js             # Landing page enhancement script
├── manifest.base.json
├── manifest.chrome.patch.json
├── manifest.firefox.patch.json
├── background/
│   ├── service-worker.js      # Message routing, badge, provider orchestration
│   └── handlers/
│       ├── badge.js           # Extension badge state management
│       ├── grammar.js         # Grammar check pipeline, token usage tracking
│       ├── settings.js        # Settings verification and status
│       └── chrome-free-ai.js  # Chrome Free AI status and download
├── content/
│   ├── writing-session.js     # Content-side typing lifecycle and stale check handling
│   ├── content.js             # Input detection, tooltip, correction adapters
│   └── content.css            # Tooltip and indicator styles
├── popup/
│   ├── popup.html             # Settings UI
│   ├── popup.js               # Settings logic
│   └── popup.css              # Popup styles
├── providers/
│   ├── abstract-provider.js  # Abstract provider contract
│   ├── abstract-openai-compatible-provider.js  # Shared logic for OpenAI-compatible APIs
│   ├── openai-provider.js     # OpenAI implementation
│   ├── chrome-free-ai-provider.js  # Chrome's built-in Gemini Nano
│   ├── ollama-provider.js          # Ollama (local LLMs)
│   ├── lmstudio-provider.js        # LM Studio (local LLMs)
│   ├── generic-openai-provider.js  # Generic OpenAI-compatible (Groq, OpenRouter, etc.)
│   └── provider-registry.js   # Provider lookup and creation
└── lib/
    ├── cache.js               # Unified TTL cache (models, availability)
    ├── config.js              # Shared configuration
    ├── logger.js              # Tagged, leveled console logger
    └── settings.js            # Settings persistence and caching

License