WebSeek

[CHI 2026 — Full Paper] Facilitating Proactive and Reactive Guidance for Decision Making on the Web: A Design Probe with WebSeek Yanwei Huang, Arpit Narechania · arXiv:2601.15100

WebSeek is a Chrome extension that helps users discover, extract, and analyze information directly on webpages. It provides an interactive canvas where users can build tables, lists, and visualizations from live web content — with an AI layer that offers both proactive suggestions (the system notices patterns and proposes next steps) and reactive assistance (users ask questions and request transformations in natural language).

An exploratory study with 15 participants revealed diverse web-analysis strategies and highlighted user preferences for transparency and control in human–AI collaboration.

---

Features

• AI-assisted data extraction — select a webpage element and let the agent extract structured rows
• Spreadsheet-style table editor — resize columns/rows, add computed columns, merge tables
• Vega-Lite visualization editor — generate and refine charts backed by table data
• Proactive AI suggestions — the extension watches your canvas and suggests next steps automatically
• Export — tables as XLSX, visualizations as SVG / PNG / JPG
• Snapshots — save and restore the full canvas state via the backend

Architecture

Chrome Extension (WXT + React + TypeScript)
   └─ Sidepanel UI  ─── REST / WebSocket ───▶  FastAPI Backend (Python)
                                                  └─ Google ADK agents
                                                  └─ OpenRouter LLM

Prerequisites

• Node.js ≥ 18
• Chrome ≥ 114 (for Side Panel API)
• The WebSeek backend running locally or on a server

Setup

1. Clone and install

git clone https://github.com/datavisards/WebSeek.git
cd WebSeek
npm install

2. Configure environment

cp .env.example .env

Edit .env:

Variable	Description
WXT_OPENROUTER_KEY	Your OpenRouter API key
VITE_BACKEND_URL	URL of the WebSeek backend, e.g. http://localhost:8000

Security: .env is gitignored. Never commit it.

Users can also enter their API key and backend URL at runtime through the extension's Settings panel (⚙ icon), without rebuilding.

3. Run in development mode

npm run dev

1. Open chrome://extensions
2. Enable Developer mode
3. Click Load unpacked → select .output/chrome-mv3-dev

4. Build for production / Chrome Web Store

npm run build

The production build is in .output/chrome-mv3/. Zip it for Web Store submission:

cd .output && zip -r webseek.zip chrome-mv3/

Project Structure

webseek/
├── entrypoints/
│   ├── background/           # Service worker
│   ├── content/              # Content script (page interaction)
│   ├── popup/                # Toolbar popup
│   └── sidepanel/            # Main UI
│       ├── components/       # React components
│       ├── apis.ts           # Backend REST calls
│       ├── macro-tools.ts    # Tool definitions for proactive AI
│       ├── macro-tool-executor.ts       # Client-side tool executor
│       ├── proactive-service-enhanced.ts # Proactive AI orchestration
│       ├── trigger-engine.ts # Rules engine for AI suggestions
│       └── prompts.ts        # LLM prompt templates
├── docs/                     # GitHub Pages tutorial
├── public/                   # Static assets
├── manifest.json             # Extension manifest (MV3)
├── wxt.config.ts             # WXT build config
└── .env.example              # Environment variable template

Security Notes

• The WXT_OPENROUTER_KEY is embedded in the extension bundle at build time. For a publicly distributed extension, route all LLM calls through your own backend so the key is never shipped to end users.
• For HTTPS pages, the backend must also be HTTPS (or use http://localhost for local development only).

Citation

If you use WebSeek in your research, please cite:

@inproceedings{huang2026webseek,
  title     = {Facilitating Proactive and Reactive Guidance for Decision Making on the Web: A Design Probe with WebSeek},
  author    = {Huang, Yanwei and Narechania, Arpit},
  booktitle = {Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems},
  year      = {2026},
  publisher = {ACM},
  url       = {https://arxiv.org/abs/2601.15100}
}

License

MIT