Policy playground

Give the agent a boundary.

Configure a policy, submit a request, and see which rule decides the outcome.

Browser simulation. No transactions are submitted. This is a model of intended behavior, not a deployed contract. All balances and addresses are illustrative.

Policy

Applied

Simulation clock

+0h 0m

Session: Active · expires in 3d 0h 0m · Time advances only when you choose.

Spent · last 24h

0 ETH

Reserved · queue

0 ETH

Available budget

2 ETH

Agent request

A simulated native-value transfer, evaluated against every active rule.

Or explore an example

Examples restart with the default policy and replace the activity log.

Decision

Ready for a request

Submit a transfer or run an example to see the policy in action.

Activity

0 requests

No requests yet. Your first decision will appear here.

What this model assumes

Only native-value transfers are modeled. Token approvals, swaps, router calldata, gas, and chain behavior are outside its scope. The spending window covers the preceding 24 hours; queued requests reserve budget until resolved. Execution rechecks the rules. Session expiry or revocation prevents pending execution.

Amounts use exact integer accounting. Nothing is persisted: refreshing starts over. These choices are illustrative and do not settle the open contract-design questions. Read the docs.