Hackathon submission · Working testnet MVP

Autonomous agents can act. Rules still hold.

RuleWallet is an open-source policy account for AI agents on Robinhood Chain testnet. It replaces unlimited wallet access with narrow roles, hard spending limits, human approval thresholds, emergency controls, and public receipts.

Run the 2-minute demo Build your testnet wallet

Submission snapshot

What judges can verify without trusting our claims.

Deployed policy contract

Dedicated scoped agent signer

Scheduled recurring strategy

Onchain simulation and enforcement

Public explorer-backed receipts

Open tests and threat model

Safety boundary: testnet ETH has no value. The contracts are unaudited. Mainnet is a separate experimental release whose autonomy fails closed until every production gate passes.

Reading Robinhood Chain testnet…

01 / PROBLEM

Agent keys are too powerful.

A useful autonomous agent needs permission to transact, but a normal wallet key can expose every asset and every contract interaction.

02 / SOLUTION

Programmable authority.

The owner defines exactly what the agent can do. The contract enforces those rules independently of prompts, models, and backend code.

03 / WHY ROBINHOOD CHAIN

Financial agents need financial rails.

An EVM-compatible chain for onchain financial infrastructure lets RuleWallet combine familiar tooling with testable automation and transparent receipts.

Architecture

Four boundaries, one verifiable flow.

Agent intent

A scheduled strategy proposes a narrow action.

Policy checks

Role, target, caps, rolling spend, nonce, expiry, and pause are verified.

Onchain execution

The exact call is simulated, signed by the scoped agent, and enforced by the contract.

Public receipt

The result is published with block and explorer evidence.

Innovation

Security is the product behavior.

The frontend explains decisions, but it cannot grant authority. The contract and role model are the source of truth.

Authority without custody

The owner keeps the admin wallet. A separate agent key receives only the narrow onchain role it needs.

Limits enforced onchain

Per-call caps, rolling spend, allowlists, nonces, expiry, and approvals remain effective if the UI or agent fails.

Simulation before signature

Every user mutation and autonomous execution is checked against current contract state before broadcast.

Public proof

Live metrics and explorer-backed receipts turn automation claims into independently verifiable evidence.

Judge flow

A complete proof in two minutes.

Start now

0:00–0:301

Inspect live guardrails

See policy state, rolling allowance, agent role, contract balance, and the observed testnet block.

0:30–1:052

Verify autonomous execution

Open the latest public receipt and compare its transaction hash with the Robinhood Chain testnet explorer.

1:05–2:003

Try to break the rules

Cross a limit or select an untrusted recipient in the simulator, then inspect the operator pause and revoke controls.

Roadmap

From testnet proof to production infrastructure.

Now

Hackathon MVP

Personal policy accounts, trusted addresses, scheduled transfers, live metrics, receipts, approvals, pause, and revoke.

Production hardening

Independent audits, stronger signer custody, monitored canary deployments, and notification delivery integrations.

Later

Mainnet gates

Independent audits, multisig administration, monitored RPC failover, canary limits, incident drills, and explicit deployment approval.

Inspect it. Break it. Verify it.

No pitch deck is required to prove the core flow. The contract, automation, receipts, tests, and failure boundaries are public.

2-minute demo

Hackathon submission · Working testnet MVP

Autonomous agents can act. Rules still hold.

Run the 2-minute demo Build your testnet wallet

Submission snapshot

What judges can verify without trusting our claims.

Deployed policy contract

Dedicated scoped agent signer

Scheduled recurring strategy

Onchain simulation and enforcement

Public explorer-backed receipts

Open tests and threat model

Safety boundary: testnet ETH has no value. The contracts are unaudited. Mainnet is a separate experimental release whose autonomy fails closed until every production gate passes.

Reading Robinhood Chain testnet…

01 / PROBLEM

Agent keys are too powerful.

A useful autonomous agent needs permission to transact, but a normal wallet key can expose every asset and every contract interaction.

02 / SOLUTION

Programmable authority.

The owner defines exactly what the agent can do. The contract enforces those rules independently of prompts, models, and backend code.

03 / WHY ROBINHOOD CHAIN

Financial agents need financial rails.

An EVM-compatible chain for onchain financial infrastructure lets RuleWallet combine familiar tooling with testable automation and transparent receipts.

Architecture

Four boundaries, one verifiable flow.

Agent intent

A scheduled strategy proposes a narrow action.

Policy checks

Role, target, caps, rolling spend, nonce, expiry, and pause are verified.

Onchain execution

The exact call is simulated, signed by the scoped agent, and enforced by the contract.

Public receipt

The result is published with block and explorer evidence.

Innovation

Security is the product behavior.

The frontend explains decisions, but it cannot grant authority. The contract and role model are the source of truth.

Authority without custody

The owner keeps the admin wallet. A separate agent key receives only the narrow onchain role it needs.

Limits enforced onchain

Per-call caps, rolling spend, allowlists, nonces, expiry, and approvals remain effective if the UI or agent fails.

Simulation before signature

Every user mutation and autonomous execution is checked against current contract state before broadcast.

Public proof

Live metrics and explorer-backed receipts turn automation claims into independently verifiable evidence.

Judge flow

A complete proof in two minutes.

Start now

0:00–0:301

Inspect live guardrails

See policy state, rolling allowance, agent role, contract balance, and the observed testnet block.

0:30–1:052

Verify autonomous execution

Open the latest public receipt and compare its transaction hash with the Robinhood Chain testnet explorer.

1:05–2:003

Try to break the rules

Cross a limit or select an untrusted recipient in the simulator, then inspect the operator pause and revoke controls.

Roadmap

From testnet proof to production infrastructure.

Now

Hackathon MVP

Personal policy accounts, trusted addresses, scheduled transfers, live metrics, receipts, approvals, pause, and revoke.

Production hardening

Independent audits, stronger signer custody, monitored canary deployments, and notification delivery integrations.

Later

Mainnet gates

Independent audits, multisig administration, monitored RPC failover, canary limits, incident drills, and explicit deployment approval.

Inspect it. Break it. Verify it.

No pitch deck is required to prove the core flow. The contract, automation, receipts, tests, and failure boundaries are public.

2-minute demo

Autonomous agents can act. Rules still hold.

Four boundaries, one verifiable flow.

Security is the product behavior.

A complete proof in two minutes.

Inspect live guardrails

Verify autonomous execution

Try to break the rules

From testnet proof to production infrastructure.

Inspect it. Break it. Verify it.

Preparing RuleWallet

Autonomous agents can act. Rules still hold.

Four boundaries, one verifiable flow.

Security is the product behavior.

A complete proof in two minutes.

Inspect live guardrails

Verify autonomous execution

Try to break the rules

From testnet proof to production infrastructure.

Inspect it. Break it. Verify it.