AI-based Translator

Translation and assistant browser extension for Chinese, English, Japanese, and Korean.

Features

• Background Execution: Translation, polishing, dictionary, and chat requests run in a background service worker. Switching tabs or closing the popup will not interrupt active operations, and finished results are automatically saved to your history.
• Translation: Translate between Chinese, English, Japanese, and Korean. Auto-detects source language.
• Text Polishing: Polish text and explain changes in Chinese.
• Dictionary: Definitions, parts of speech, root/affix analysis, and bilingual example sentences.
• Chat: Sidebar chat assistant with conversation memory.
• History: Saves the last 10 translation records locally.

Installation

From Microsoft Edge Add-ons Store

1. Click the badge above to open the Edge Add-ons store page.
2. Click Get to install.

Sideload (Developer Mode)

1. Clone this repository:

   git clone https://github.com/OpenTester007/Chat-with-KB.git

2. Open edge://extensions/ in Microsoft Edge.
3. Enable Developer mode.
4. Click Load unpacked and select the cloned folder.
5. Open the extension Settings tab and configure your API key.

Configuration

Setting	Default	Description
API Key	empty	Add your own NVIDIA Build or OpenAI-compatible provider key.
API Endpoint	https://integrate.api.nvidia.com/v1	Customize for OpenAI-compatible providers. Remote endpoints must use HTTPS.
Model	openai/gpt-oss-20b	Choose from presets or enter any model ID.
Streaming	On	Toggle between real-time streaming and batch responses.

Supported Models (Presets)

Model	Type
openai/gpt-oss-20b	Reasoning (default)
deepseek-ai/deepseek-v4-flash	Chat
meta/llama-3.3-70b-instruct	Chat
google/gemma-4-31b-it	Chat

You can also enter any custom OpenAI-compatible model ID.

Using Local Ollama

To use a local Ollama instance, set the environment variable and configure the endpoint:

set OLLAMA_ORIGINS=*
ollama serve

Then in the extension settings, set:

• API Endpoint: http://localhost:11434/v1
• Model: your Ollama model name, such as llama3, qwen2.5, or mistral
• API Key: leave empty for local endpoints that do not require a key

Remote non-local HTTP endpoints are intentionally blocked; use HTTPS for hosted providers.

Tech Stack

• Manifest V3
• Vanilla JavaScript
• OpenAI-compatible Chat Completions API

Development

git clone https://github.com/OpenTester007/Chat-with-KB.git
cd Chat-with-KB
# Load unpacked in edge://extensions/ (Developer mode)

Edit popup.js, popup.html, background.js, or style.css, then reload the extension from edge://extensions/.

Basic checks:

node --check popup.js
node --check background.js

Project Structure

manifest.json          # Extension manifest (MV3)
background.js          # Service worker and API request handling
popup.html             # Main popup UI
popup.js               # Popup UI logic
style.css              # Styles
screenshots/           # Screenshots for documentation
16.png / 32.png        # Extension icons
48.png / 128.png       # Extension icons
LICENSE                # MIT License
CONTRIBUTING.md        # Contribution guidelines
CODE_OF_CONDUCT.md     # Community code of conduct
ROADMAP.md             # Planned features
CHANGELOG.md           # Release history

Privacy

Settings, chat history, and translation history are stored locally with the browser storage API. Text entered by the user is sent to the configured API provider for processing.

License

MIT, see LICENSE.