$10 daily cap.
$1 per-tx ceiling.
Recipient whitelist with TTLs.
All enforced by a smart contract. Not the backend. Not the prompt.
The agent gets a wallet. The wallet refuses to drain.
Giving your AI agent the wallet key is like giving your toddler the credit card.
They'll still buy 200 stuffed animals.
Spend caps belong on chain. Not in the prompt.
Give your AI an allowance.
Set a daily budget, approve who it can pay. Anything else, the smart contract refuses โ even if the agent gets hacked or hallucinates.
Alpha live on Solana devnet. No real funds.
@VaultMindAI Full report on Enclz: https://t.co/Wwa1QoZYuZ
An audit isn't a finish line. It's a feedback loop.
We'll keep running it as we approach mainnet.
We just ran @VaultMindAI on Enclz.
Score: 100/100 on the Solana program that will hold real funds under user policy.
A short thread on why continuous AI security is becoming the new floor for smart contracts.
@VaultMindAI What that looks like in practice:
You push a change. The scanner runs against the new code in seconds. You see the deltas โ what got safer, what got riskier โ before the branch merges.
The audit doesn't end. It moves with the code.
We are on-chain spend policy for AI agents on Solana. Limits live in an Anchor program, so even a backend compromise can't drain wallets.
Scaffolded on @eitherway. RPC by @QuickNode, auth via @solflare_wallet.
Submitted for @colosseum Frontier hackathon ๐จโ๐ป
It's the final day of the @Solana Frontier Hackathon.
Thousands of developers and founders from across the world are now submitting the products they have been building over the past couple of months.
Remember, today is not the end of your journey. It's just the beginning.
Why we built Enclz? Because:
https://t.co/OYXeYmXxWv and @agentcashdev have no policies
@OpenWallet policies are agent-reachable
@openfort_hq requires SDK integration
@CoinbaseDev wallets are centralized
We are @solana native, trustless and ready for production use ๐
72 live x402-gated APIs in week one. That's the rail working.
The wallet underneath is what makes it safe. Per-tx caps, recipient whitelists, on-chain enforcement.
A compromised agent can't drain those API budgets because the contract won't let it.
AI agents will do 99% of on-chain transactions soon.
They need payment rails to pay for tasks โ x402
They need safety net to prevent hacks โ ๐ฅ Enclz
We provide payments safety for autonomous world
Pay-per-request changes the threat model for agent wallets.
One prompt injection isn't one bad transaction anymore. It's a thousand pay-per-request calls drained before anyone notices.
Rate-limit at the contract. Not at the agent.