Definition

Oracle

A service that brings off-chain data (prices, real-world events, weather, election outcomes) onto a blockchain so smart contracts can use it.

Smart contracts cannot natively access data outside their blockchain — they only see what's already on-chain. An oracle bridges this gap. The most common use: price feeds. A DeFi lending protocol needs to know the price of ETH to determine when a borrower's collateral is liquidatable. Chainlink aggregates prices from many sources and writes the result on-chain for contracts to query.

Oracle design is hard. A single oracle creates a single point of failure — manipulating one source can drain a protocol. Decentralized oracles aggregate many sources, with mechanisms to reject outliers. Even so, oracle manipulation has caused numerous DeFi exploits (notably bZx, Cream, Mango Markets) where attackers spiked or crashed a price feed for one block to game lending positions.

Beyond prices, oracles can deliver any verifiable off-chain data: sports scores, weather, randomness (verified via VRF), election outcomes. UMA pioneered "optimistic oracles" where data is posted with a dispute window rather than active aggregation — cheaper but slower.

Why it matters

Oracles are the security boundary between DeFi protocols and the real world. A protocol's safety is bounded by its oracle's integrity.

How CryptoRadar24 tracks it

CryptoRadar24 references oracle issues when reporting on DeFi protocol exploits or specific market-data anomalies.

Related terms

FAQ

What is the most popular oracle?

Chainlink dominates — it secures the majority of DeFi TVL via decentralized price feeds. Other notable oracles include Pyth (Solana-native, low-latency), API3, Band Protocol.

Why are oracle attacks common?

Smart contracts trust oracle output unconditionally. If an attacker can manipulate the price an oracle reports for one block, they can drain pools that price collateral against that feed.

Are oracles centralized?

Decentralized oracles (Chainlink) aggregate many independent nodes that submit data; the median result is reported. Centralized oracles (a single source) are cheaper but trust-heavy.

Can oracles be wrong?

Yes — both via manipulation (attacker influencing input sources) and via bug (formatting issues, data feed errors). Aggregator design tries to filter outliers but is not perfect.