Skip to main content

​
Core Concepts

Understanding how Extensy works helps you write better prompts, interpret generated code, and debug issues faster. This page covers the key ideas behind the platform.

​
The generation pipeline

When you submit a prompt, Extensy runs it through a five-stage pipeline:
1

User prompt

You describe your extension’s behavior in plain English. The more specific you are about the target site, the action, and any UI, the better the output.
2

AI analysis

Extensy’s proprietary LLMs — trained on millions of lines of Chromium API code — parse your prompt and map your intent to the correct Chrome APIs, permission scopes, and script contexts.
3

Structured generation

The AI generates all required files: manifest.json, background service worker, content scripts, and popup UI. Each file is generated with awareness of the others to ensure they work together.
4

Validation

The output is checked for common syntax errors and verified against the Manifest V3 specification before it reaches you.
5

ZIP assembly

All files — including icons at the required sizes — are packaged into a single ZIP archive ready to load in Chrome or submit to the Chrome Web Store.

​
Manifest V3 compliance

Every extension Extensy generates follows the Manifest V3 (MV3) standard — the current, required format for all Chrome extensions. MV3 improves security, performance, and user privacy compared to the legacy Manifest V2.
Extensy handles all the complexity of MV3 service workers and declarative network rules automatically. You don’t need to configure these manually.
The Chrome Web Store requires MV3 for all new submissions and updates. Extensions still using MV2 will eventually stop working in Chrome.

​
Anatomy of a generated extension

When you unzip your downloaded package, you’ll find the following files:
The root configuration file for your extension. It tells Chrome the extension’s name, version, description, required permissions, and which scripts to load in which contexts. Every other file in the package is referenced from here.
A script that runs in the context of webpages the user visits. Use content scripts to read or modify the DOM, extract data from a page, or inject UI elements. Content scripts run in an isolated world — they share the page’s DOM but not its JavaScript globals.
A service worker that listens for browser-level events such as tab updates, navigation events, or messages from content scripts and the popup. Because it is event-driven, it does not run continuously — design it to respond to events and terminate gracefully.
A folder containing your extension’s icon at three required sizes: 16px, 48px, and 128px. Chrome uses these in the toolbar, the extensions management page, and the Chrome Web Store listing.

​
Credit system

Extensy uses a credit-based model for generation.

One generation = one credit

Each time you click Generate and receive a complete ZIP, one credit is deducted from your balance.

Regeneration costs a credit

Refining your prompt and clicking Regenerate starts a new generation and deducts an additional credit.

Failed generations are refunded

If a generation fails due to a server-side error, your credit is returned automatically. You are never charged for a failure on Extensy’s end.

​
Iterative development

The most effective way to build with Extensy is to iterate in small steps. Start with a minimal prompt that captures the core behavior, load and test the result in Chrome, then return to add more features or fix issues with a refined prompt.
Avoid writing a single long prompt that tries to describe every feature at once. Build the core behavior first, verify it works, then extend it in follow-up generations.