AI agents that can’t overspend.

Caps, allowlists, and human approval on every payment — provably enforced on Hedera.

Trust us as little as you like.

L0Off-chain

Off-chain rules

Caps, an allowlist, and approvals — checked before every payment runs.

Stops a misbehaving agent.

L1Verifiable

On-chain audit

Every decision and rule change is recorded on Hedera. Anyone can re-check the numbers.

The history can’t be faked.

L2Trustless

On-chain enforcement

A smart contract enforces the limits itself.

Even a hacked agent can’t overspend.

+ Non-custodialFunds never leave your wallet — you sign every payment yourself.
Drop-in npm package

Govern any agent in two lines.

One call sets up the topics, storage, and policy, and hands you Agent Kit hooks. Or do it from the terminal — zero code.

agent.ts
import { createCurb } from 'hedera-curb';

const curb = await createCurb({ client, agentAccountId });
new HederaAIToolkit({ client, configuration: {
  context: { hooks: curb.hooks },   // ← governed
}});

// or, no code:
//   $ npx hedera-curb init --network testnet
VERIFIABLE SPEND-CONTROL   ·   ENFORCED ON-CHAIN   ·   AGENTS CAN PAY — NOW THEY CAN’T OVERSPEND   ·   CURB × HEDERA   ·   VERIFIABLE SPEND-CONTROL   ·   ENFORCED ON-CHAIN   ·   AGENTS CAN PAY — NOW THEY CAN’T OVERSPEND   ·   CURB × HEDERA   ·