upzare/Waffy

upzare/Waffy has 43 GitHub stars and 2 forks. GitHot measured 0 stars in 24 hours and 0 in 7 days. Last checked 2026-11-09 21:58 UTC. Last push 2026-11-02.

How these counts are measured

Waffy - AI Copilot for Web

43 stars 2 forks

What you can do

Natural language automation — Tell Waffy what to do in plain English and it plans and executes browser actions for you.
Vision-based interaction — Uses screenshots and coordinates to click, type, scroll, and navigate like a human would.
Multi-stage pipeline — Separate models for planning, execution, validation, and output so each step can use the right model for the job.
Browser built-in AI — Optional on-device models (Chrome Gemini Nano / Edge Phi Mini) for lightweight tasks like title and step generation. No API key required.

Measured star growth

in 24h: 0
in 7 days: 0
in 15 days: Collecting history
in 1 month: Collecting history

Net star changes over overlapping periods, not daily totals. Missing measurements are left blank. Stars measure attention, not software quality.

README.md

Original

Waffy

Waffy is an open-source browser extension that brings AI-powered browser automation to your side panel. Describe what you want in natural language — Waffy can browse pages, click elements, fill forms, extract data, and complete multi-step workflows using vision-capable models and tool calling.

Features

Natural language automation — Tell Waffy what to do in plain English and it plans and executes browser actions for you.
Vision-based interaction — Uses screenshots and coordinates to click, type, scroll, and navigate like a human would.
Multi-stage pipeline — Separate models for planning, execution, validation, and output so each step can use the right model for the job.
Browser built-in AI — Optional on-device models (Chrome Gemini Nano / Edge Phi Mini) for lightweight tasks like title and step generation. No API key required.
Bring your own API keys — Connect your own provider accounts. Keys are stored locally in the extension — Waffy does not run a backend for inference.
Multiple providers — OpenAI, Anthropic, Google AI Studio, xAI, Groq, OpenRouter, and Browser built-in AI.
Flexible model configuration — Choose different models for each pipeline stage and UI helper task in settings.
Chat history & pinned prompts — Keep conversation history and customize quick-start prompts on the home screen.
Side panel UI — Open Waffy from the toolbar or with Ctrl+I (configurable in chrome://extensions/shortcuts).

Supported providers

Provider	Example models
OpenAI	GPT-4o, GPT-4.1
Anthropic	Claude Sonnet, Claude Opus
Google AI Studio	Gemini 2.x
xAI	Grok
Groq	Llama, Mixtral
OpenRouter	100+ models via a single API key

Add API keys in Extension settings → API Keys after installation. For browser built-in AI, download the on-device model from Settings → Models.

Choosing models

Waffy runs a multi-stage pipeline. Each stage can use a different model:

Stage	Purpose	Model guidance
Planning	Decides whether to automate and drafts a task plan	Fast text models work well
Execution	Drives browser actions from screenshots	Use a vision model with spatial reasoning and image grounding — it must identify UI element coordinates on screenshots. Recommended: gemini-3.6-flash (Google AI Studio or OpenRouter). Browser built-in AI is not recommended here.
Validation	Checks whether the task succeeded	Fast text models work well
Output	Summarizes results for you	Fast text models work well
Title / Step	Short UI labels	Browser built-in AI works well (default)

Getting started

Prerequisites

Bun 1.0+
A Chromium-based browser (Chrome, Brave, Edge, Opera, etc.)

Build from source

git clone https://github.com/upzare/Waffy.git
cd Waffy
bun install
bun run build

This produces a build/ directory and a .output/Waffy-1.0.0.zip package.

Load the extension

Open the extensions page for your browser:
Enable Developer mode
Click Load unpacked and select the build/ folder
Click the Waffy icon in the toolbar, or press Ctrl+I to open the side panel

Configure API keys

Right-click the Waffy icon → Options, or open Extension settings from the side panel
Go to API Keys and add a key for at least one provider
Go to Models and pick models for each pipeline stage. For Execution, use a vision model with spatial reasoning — e.g. gemini-3.6-flash via Google or OpenRouter.
Start chatting — try something like "Go to example.com and click the login button"

Development

bun run dev    # watch mode — rebuilds on file changes
bun run start  # dev server with hot reload

After bun run dev, reload the extension in chrome://extensions/ when you make changes.

Project structure

├── extension/
│   ├── src/
│   │   ├── panel/          # Side panel UI (chat, history, input)
│   │   ├── settings/       # Extension options page (API keys, models, general)
│   │   └── lib/
│   │       ├── llm/        # Model providers, streaming, prompts, tools
│   │       ├── agent.ts    # Entry point for AI calls
│   │       ├── client.ts   # Chrome storage & settings
│   │       ├── background.ts
│   │       └── content.ts  # Page interaction (clicks, typing, screenshots)
│   ├── public/             # Manifest, icons, static assets
│   ├── package.json
│   ├── webpack.config.ts
│   └── tsconfig.json
├── package.json            # Workspace root — run bun commands from here
├── build/                  # Production output (load this in Chrome)
└── .output/                # Zipped release packages

Privacy

API keys and conversation history are stored locally in your browser via Chrome's extension storage. Requests go directly from your browser to the AI provider you configure — Waffy does not proxy or log your prompts.

See waffy.io/privacy for the full privacy policy.

Contributing

Contributions are welcome. Please read CONTRIBUTING.md before opening a pull request.

This project follows the Contributor Covenant Code of Conduct.

Security

To report a security vulnerability, see SECURITY.md. Please do not open public issues for security reports.

License

Licensed under the Apache License, Version 2.0. See LICENSE for the full text.