Turn any API docs into a live-tested SDK. Instantly.

An agentic multi-agent LLM system that eliminates the integration tax.

> Reading documentation..._
View on GitHub
Python 3.11+ LangGraph Gemma 4 on GCP FastAPI

The Integration Tax

Integrating a new API takes 3โ€“7 hours of manual pain.

1. Reading Docs

Endless scrolling through poorly structured, outdated documentation just to find the right endpoint.

2. Boilerplate

Writing the same HTTP client, error handling, and type definitions over and over again.

3. Debugging

Wasting hours figuring out why an undocumented field is breaking your implementation.

How It Works

1. URL Provide API Docs
โ†’
2. Supervisor Orchestrates Agents
โ†’
3. The 6 Agents Research & Build
โ†’
4. VS Code Typed SDK Ready

Meet The 6 Agents

๐Ÿ‘”

Supervisor

Coordinates the multi-agent workflow and handles routing.

๐Ÿ”

Researcher

Scrapes and understands the API documentation deeply.

๐Ÿ“

Architect

Designs the SDK structure, models, and type definitions.

๐Ÿ’ป

Engineer

Writes the actual code for the SDK client and methods.

๐Ÿงช

QA Tester

Runs live HTTP tests to verify the generated SDK works.

๐Ÿ“ฆ

Packager

Bundles the code, generates READMEs, and prepares distribution.

Built With

Backend Python, FastAPI, Pydantic
AI / Agentic LangGraph, Gemma 4 on GCP
Scraping Playwright, BeautifulSoup
Testing Pytest, HTTPX
Voice ElevenLabs, WebSockets
Frontend / Bridge Vite, React, Chrome Extension

See SDKGen in Action

โ–ถ
Demo Video Placeholder

From raw docs URL to working SDK in under 60 seconds.

Tested On Tough APIs

Stripe API

Challenge: Massive API surface, deeply nested types.

Result: Fully typed SDK with complex pagination handled.

OpenWeatherMap

Challenge: Inconsistent error formats, varying unit params.

Result: Standardized error handling and unified units.

Legacy Finance API

Challenge: XML responses, terrible documentation.

Result: Seamless XML to JSON parsing, hidden endpoints mapped.

Quickstart

1. Backend Setup

git clone https://github.com/ayanokojix21/SDKGen
cd SDKGen
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
uvicorn backend.main:app --reload

2. Chrome Extension

cd chrome-extension
npm install
npm run build
# Load unpacked extension in Chrome
# pointing to /dist

3. VS Code Extension

cd vscode-extension
npm install
npm run compile
# Press F5 to start debugging
# Use Command Palette to run SDKGen