An oracle solves what blockchain engineers call the "oracle problem": a smart contract can only see data that already lives on its own chain, yet most useful contracts need to react to something happening outside it, a stock price, a flight delay, a weather reading, or a transfer on another network. An oracle is the middleware that fetches that external information, checks it, and delivers it on-chain in a form every validator can independently verify, since a contract cannot simply make its own HTTPS request without breaking network consensus.

Oracles work in one of two main patterns. Push oracles publish fresh data on-chain at fixed intervals or whenever a value moves past a set threshold, so it is already sitting on the blockchain when a contract needs it. Pull oracles instead sign data off-chain and let a contract fetch it on demand, which lowers costs for feeds that update far more often than they are actually read. Sources range from software, APIs, and exchanges to hardware sensors reporting real-world conditions like location or temperature.

Because a single data provider can be bribed, hacked, or go offline, serious DeFi protocols rely on decentralized oracle networks that combine independent nodes and aggregate their results, reducing the risk that any one source can move a price. Chainlink is the largest of these, securing tens of billions of dollars across lending markets, stablecoins, and derivatives, while Pyth and RedStone offer faster, pull-based feeds built for high-frequency trading. When a feed is wrong or manipulated, the consequences are direct: lending platforms have suffered real losses from bad price data triggering incorrect liquidations, which is why oracle reliability is treated as core DeFi infrastructure rather than a minor technical detail.