A decentralized keeper is an autonomous agent – or a network of such agents – that monitors on-chain conditions and executes smart contract functions when those conditions are met. Keepers handle the “cron jobs” of DeFi: liquidations, oracle price updates, reward harvesting, limit-order execution, and any other task that a contract cannot trigger by itself because the EVM has no built-in scheduler.

Why Keepers Exist#

Smart contracts on Ethereum and other chains are reactive – they run only when an external account submits a transaction. Many protocol-critical actions (liquidating an under-collateralized loan, settling an expired option, compounding yield-farming rewards) must happen promptly, yet the contract cannot call itself. Keepers bridge this gap by watching for trigger conditions off-chain and submitting the necessary transaction when the time comes.

Why Decentralization Matters#

A single centralized bot can perform the same mechanical task, but it introduces a single point of failure. If the operator goes offline, misses a block, or acts maliciously, the protocol suffers. Decentralized keeper networks distribute the work across many independent participants, which provides:

  • Liveness – if one keeper is down, others pick up the task.
  • Censorship resistance – no single party can withhold execution.
  • Economic alignment – keepers compete for rewards, which keeps execution prompt and costs competitive.

Common Use Cases#

Task Example protocols
Liquidations Aave, Compound, Maker
Oracle price-feed updates Chainlink data feeds
Yield harvesting and auto-compounding Yearn, Beefy
Limit and stop-loss order execution DEX aggregators, on-chain order books
Rebalancing LP positions Concentrated-liquidity managers
Options and futures settlement On-chain derivatives protocols

Avoiding Duplicate Execution#

When multiple keepers watch the same trigger, naive execution would waste gas on redundant transactions. Networks use several techniques to prevent this:

  1. On-chain state locks – the contract flips a flag or advances a nonce as part of the keeper’s transaction, causing any competing transaction to revert.
  2. Turn-based rotation – keepers are assigned time slots or round-robin order so only one is eligible at any moment. Chainlink Automation uses a rotating-leader model.
  3. First-past-the-post rewards – only the transaction that actually changes state earns the reward; duplicate attempts simply burn gas, which is a strong economic disincentive.
  4. Off-chain coordination – some networks let keepers signal intent before submitting, reducing on-chain collisions.

Major Implementations#

Chainlink Automation is the most widely adopted keeper service. Developers register an “upkeep” by deploying a contract that implements two functions: checkUpkeep (a view function that returns whether work is needed) and performUpkeep (the state-changing call). Chainlink’s decentralized node network monitors checkUpkeep off-chain and calls performUpkeep when it returns true. Upkeeps are funded with LINK tokens, and the network uses a rotating-leader design to avoid duplicate execution.

Gelato Network#

Gelato offers a flexible automation layer that supports time-based triggers, on-chain condition checks, and off-chain “resolver” functions written in TypeScript. It supports multiple chains and integrates tightly with smart-contract wallets. Tasks can be funded with the protocol’s native token or with the network’s fee token.

Keep3r Network#

Created by Andre Cronje, Keep3r is a more permissionless marketplace: any protocol can post a “job,” and any bonded keeper can execute it. Keepers earn KP3R tokens for completed work. The system relies on bonding and slashing to deter misbehavior rather than on a managed node set.

Risks and Limitations#

Gas-cost spikes – during periods of extreme network congestion, keeper rewards may not cover gas costs, causing delays in time-sensitive actions like liquidations. This is a liveness risk for any protocol that depends on prompt keeper execution.

Incentive misalignment – if rewards are too low relative to gas costs or competing MEV opportunities, keepers may deprioritize a protocol’s tasks. Some protocols mitigate this by offering dynamic, gas-adjusted rewards.

MEV extraction – keepers that perform liquidations or large swaps can be sandwiched or front-run by MEV searchers, reducing their effective reward and potentially harming the protocol’s users.

Smart-contract risk – keepers interact with protocol contracts, so a vulnerability in either the keeper infrastructure or the target contract can be exploited. Both sides need thorough auditing.

Oracle dependency – keepers that trigger based on price data inherit the risks of the underlying oracle. Stale or manipulated feeds can cause incorrect liquidations or settlements.