SAFETY RAILS FOR
ON-CHAIN AI AGENTS
Add guardrails to any AI agent that touches blockchain. Validate transactions before signing.
Verified on BaseRecent Activity
Live validation feed across all chains.
See It In Action
Watch ProofGate block an infinite approval in real-time.
Live demo • ProofGate validates every transaction before execution
Process
A deterministic workflow for transaction safety.
Define Guardrail
Set safety rules for your AI agent: balance thresholds, spending limits, allowed contracts. Simple or custom constraints.
Validate
Before signing any transaction, ProofGate validates it against your guardrail. Balance checks, contract whitelists, spending limits.
Execute
Receive SAFE/BLOCKED verdict with cryptographic proof recorded on-chain. Only execute transactions that pass your safety rules.
Why You Need Guardrails
AI agents are powerful. They're also unpredictable.
The Problem: AI Agents Can't Be Fully Trusted
Prompt Injection: An attacker tricks your agent into sending all funds to their address by injecting malicious instructions into user input.
Hallucinations: Your trading bot "thinks" it's buying 10 tokens but accidentally approves unlimited spending on a malicious contract.
Logic Errors: A bug in your agent's code causes it to execute a $10,000 trade when you meant $100.
❌ WITHOUT GUARDRAILS
- • Agent drains entire wallet in one bad trade
- • Unlimited approvals to unknown contracts
- • No way to prove what went wrong
- • Funds lost = gone forever
✅ WITH GUARDRAILS
- • Max spending limit per transaction ($100, $1k, etc.)
- • Whitelist of approved contracts only
- • Balance requirements before executing
- • On-chain proof of every validation
Think of Guardrails as Your Agent's Safety Rails
You wouldn't give a self-driving car unlimited speed with no brakes. Same principle applies to AI agents with wallet access. Guardrails let you:
- Define boundaries: "Only trade on Uniswap, max $500 per transaction"
- Prevent disasters: No single mistake can drain your wallet
- Build trust: Cryptographic proof of compliance for audits
Use Cases
Guardrails for any AI agent touching blockchain.
DeFi TRADING BOTS
Enforce max trade size, slippage limits, and approved DEXs. Prevent one bad trade from draining the entire wallet.
NFT PURCHASE AGENTS
Validate collection authenticity and floor prices before buying. Stop agents from overpaying or purchasing fakes.
TOKEN TRANSFERS
Set spending limits and whitelist recipient addresses. Prevent accidental or malicious large transfers.
TREASURY MANAGERS
Validate multi-sig proposals before signing. Enforce treasury guardrails with cryptographic proof for compliance.
Security Stack
Every validation runs all checks automatically. No configuration needed.
Integrations
Drop-in plugins for popular AI agent frameworks.
TypeScript SDK
Core SDK for any TypeScript/JavaScript project. Full validation API.
npm i @proofgate/sdkEliza Plugin
Drop-in plugin for ai16z Eliza framework. Auto-validates all transactions.
npm i @proofgate/eliza-pluginGOAT Plugin
Guardrails for GOAT SDK. Protect 50+ DeFi plugins (Uniswap, Aave, etc).
npm i @proofgate/goat-pluginFAQ
Technical details about the validation process.