​

Architecture

PolicyKit is designed as a layered system where each component has a clear responsibility. This page explains how the pieces fit together.

​

System Overview

​

Component Responsibilities

​

SDK Layer

The SDK provides the developer-facing API:

​

Smart Contract Layer

The on-chain contracts enforce policies at the transaction level:

​

Off-Chain Layer

Lit Protocol handles rules that require external data or complex computation:

​

Data Flow

​

Policy Deployment

1. Developer builds a policy with PolicyBuilder
2. Full policy JSON is pinned to IPFS via IPFSClient
3. On-chain rules are encoded via PolicyEncoder
4. PolicyEngine.setPolicySet() is called with:
   • Policy CID (as bytes32)
   • PKP address (for attestation verification)
   • Encoded on-chain rules
   • Fail mode and attestation requirements

​

Transaction Evaluation

1. A transaction is submitted to the smart account
2. PolicyGuard or PolicyKit7579Module intercepts it
3. On-chain rules (Tier 1 and 2) are evaluated by the PolicyEngine
4. If off-chain rules exist, a Lit Action is triggered
5. The Lit Action fetches the policy from IPFS using the CID
6. Off-chain rules (Tier 3) are evaluated
7. If all rules pass, the Lit network threshold-signs an EIP-712 attestation
8. The attestation is verified on-chain by AttestationVerifier
9. Transaction proceeds only if all checks pass

​

Security Model

​

Trust Assumptions

• On-chain rules: Fully trustless — evaluated by the EVM
• Off-chain rules: Trust the Lit Protocol network’s decentralized TEE infrastructure
• IPFS storage: Content-addressed — CID ensures policy integrity
• Attestations: EIP-712 typed signatures verified on-chain

​

Attack Surface

PolicyKit minimizes the attack surface by:

• Keeping critical rules on-chain where possible
• Using threshold cryptography (no single point of failure for attestations)
• Content-addressing policies on IPFS (tamper-evident)
• Supporting “closed” fail mode for high-security deployments