Built for The Synthesis Hackathon 2026

On-Chain Rules
for AI Agents

Define exactly what your AI agent can spend, who it can interact with, and what it can reveal. Every action verifiable on Base. You stay in control.

Launch DashboardView on GitHub
Base L2
Chain
8/8
Tests
ERC-8004
Standard
Solidity
Language
agentwill.config.ts
const ruleSet = {
spending: { dailyLimit: "$1,000", perTx: "$100" },
interactions: ["UniswapRouter", "Aerodrome"],
privacy: "strict"
}
await
agentWill.deploy(ruleSet)
Rule set deployed on Base
The Problem

Agents have blank cheque access

When you give an agent access to your wallet, there is nothing enforcing what it can and cannot do.

No Spending Limits

Your agent can drain your entire wallet. No per-transaction caps, no daily limits, no boundaries.

No Interaction Controls

Agents call any contract, any function. No way to whitelist or blacklist specific protocols.

No Privacy

Every API call leaks your spending patterns, contacts, and behavior. You never consented to that exposure.

Humans are either forced to babysit their agents constantly, or hand over control entirely and hope for the best.

Neither is acceptable as agents become more capable and autonomous.

Features

Everything an agent needs to stay accountable

AgentWill is the single infrastructure layer addressing spending, interaction, privacy, and accountability — all at once.

CORE

Spending Controls

Per-token daily limits and per-transaction caps. Agent spends freely within your boundaries.

Interaction Rules

Whitelist or blacklist contracts and function selectors. Full scope control over agent behavior.

Privacy Controls

Choose what metadata your agent reveals on-chain. Keep patterns and behavior private.

On-Chain Verification

Every action verified against smart contract before execution. All logs immutable on Base.

ERC-8004

ERC-8004 Identity

On-chain agent identity built for the ERC-8004 standard. Real identity, portable everywhere.

Activity Dashboard

Real-time monitoring of every agent action. See what was allowed, blocked, and why.

Before vs After

Before AgentWill vs After AgentWill

Without AgentWill
The old way
Agent has unlimited wallet access
No spending controls or limits
Any contract can be called
All activity leaks metadata
No way to verify agent behavior
Human must babysit constantly
With AgentWill
The right way
Agent operates within defined rules
Per-token daily & per-tx limits
Only whitelisted contracts callable
Configurable privacy controls
Every action logged on-chain
Human stays in control, passively
How it Works

Simple by design

From rule definition to on-chain enforcement in minutes.

01

Define Your Rules

Set spending limits, interaction permissions, and privacy controls through a clean interface. No code required.

02

Deploy On-Chain

Rules compile to an immutable smart contract on Base. Survives any platform, any provider.

03

Agent Operates

Agent queries the contract before each action. Contract allows or blocks — permanently logged.

04

You Stay in Control

Monitor activity in real-time. Update rules anytime. Revoke access instantly if needed.

Base Sepolia
Chain
8/8
Tests Passing
2
Smart Contracts
ERC-8004
Agent Standard
Interactive Demo

Create your first rule set

Define spending limits and interaction permissions. See how AgentWill enforces boundaries automatically.

Spending Rules

Limits per token

Reveal Metadata

Make activity publicly visible

Frequently Asked Questions

Everything you need to know about AgentWill.

Built by Shikhar Verma

Student, builder, hackathon participant

Shikhar Verma

@Shikhyyy

Building at the intersection of AI agents and on-chain infrastructure. Participating in The Synthesis Hackathon 2026.

Follow on XGitHub
The Synthesis Hackathon 2026

Submitted to 5 Tracks

Agents With Receipts — ERC-8004

$4,000

Protocol Labs

stETH Agent Treasury

$3,000

Lido Labs

Agent Services on Base

$5,000

Coinbase

Best Use of Delegations

TBD

MetaMask

Synthesis Open Track

Funded

Community