IO Saga: Real-Time CS2 Skin Economy

What it is

IO Saga is a live economy platform built around Counter-Strike 2 skins. Prices come from real market data, you buy instantly, and the item reaches you through an automated Steam trade offer. Nobody has to handle anything manually or wait for a human to hit accept.

Architecture

The backend is Node.js and Express with a raw PostgreSQL layer. I skipped the ORM on purpose, because explicit SQL keeps the trade-critical queries auditable. Redis handles sessions, rate limiting and hot price caches, and Socket.io pushes live price moves and trade status to the client.

The Steam layer is the hard part. A pool of steam-user bots holds authenticated sessions, with tokens from steam-totp, and steam-tradeoffer-manager runs the whole offer lifecycle: proposal, sent, awaiting confirmation, accepted, completed. Mobile confirmations are automated through the Steam confirmation protocol, and every state a trade can get stuck in has its own retry logic and error recovery.

The data side

Prices don’t come from Valve. They come from the secondary market. A continuous ingestion pipeline (see the Market Price Pipeline) feeds Buff163 market data into Postgres, and the platform recalculates item values, trends and spread signals on every ingest. That’s what makes instant buy and sell pricing possible without taking on inventory risk.

What I own

All of it: schema design, trade engine, bot infrastructure, the real-time layer, UI, deployment and monitoring. The interesting problems are distributed-systems problems. Settling trades idempotently, recovering from a crash in the middle of a confirmation, and keeping three independent sources of price truth consistent when something fails.