Agentic Proof-Chaining Framework

An On-Chain Framework for Verifiable AI Collaboration

---

🚀 Live Interactive Demo

Experience the framework live, without installing anything. The community validator dashboard is deployed via GitHub Pages and allows you to interact with the deployed contracts directly from your browser.

➡️ Launch the Live Dashboard

---

Overview

This project provides the on-chain infrastructure for verifiable, multi-agent AI collaboration. It enables autonomous AI agents to prove the integrity of their computational work using zero-knowledge proofs, and to chain these proofs together to perform complex, multi-step tasks.

The core of the framework is a suite of smart contracts that can verify ZK proofs generated by off-chain agents, and an orchestration layer that manages the delegation of tasks and the composition of proofs.

---

Live Deployment (Sepolia Testnet)

The core infrastructure is live, deployed, and verified on the Ethereum Sepolia testnet.

Contract	Live Sepolia Address	Etherscan Link
RWKV Verifier	0x52b5e61fA6Ae53BA08B9094eA077820283Dcec01	View on Etherscan
Mamba Verifier	0x89dFdcC74Ed05bf2a76eF788b15e5cbC8Ad8C5cD	View on Etherscan
xLSTM Verifier	0x52a55dEBE04124376841dF391Ef0e4eF1dd6835B	View on Etherscan
Agentic Orchestrator	(See maximal_deployment_addresses.json)	(Deploy with 'make deploy-sepolia')

---

Architecture

The framework is divided into two main components: the off-chain agents and the on-chain contracts.

graph TD
    subgraph "Off-Chain Environment"
        A[AI Agent] -- 1. Performs Computation --> B(ML Model)
        B -- 2. Generates ZK Proof --> C((ZK Proof))
    end

    subgraph "On-Chain Environment (Ethereum)"
        C -- 3. Submits Proof --> D{Verifier Contract}
        D -- 4. Verifies Proof --> E[Verification Result]
        D -- 5. Mints Receipt --> F(Computation Receipt)
        F -- 6. Informs --> G[Orchestrator Contract]
    end

    style D fill:#e1f5fe,stroke:#333,stroke-width:2px
    style G fill:#e3f2fd,stroke:#333,stroke-width:2px

Loading

1. Off-Chain Agents: Responsible for performing computationally intensive tasks (e.g., running an ML model) and generating a ZK proof of that computation using EZKL.
2. On-Chain Verifiers: A suite of Solidity contracts, each specialized to a particular AI architecture, that can verify the correctness of a ZK proof.
3. On-Chain Orchestrator: A central contract that manages the delegation of tasks to agents and the composition of their proofs into a verifiable chain.

---

Visualizations & Core Flow

To better understand the flow of data and control through the system, please see the following resources:

• docs/CORE_VERIFIER_FLOW.md: A detailed markdown document that explains the step-by-step process of verification.
• scripts/ascii_verifier_flow.py: A Python script that generates a simple ASCII art representation of the verifier flow.
• scripts/verifier_flow_visualization.py: A Python script that generates a more detailed, graphical visualization of the verifier flow.

---

End-to-End Demo

To run a complete, end-to-end test of the framework (from proof generation to on-chain verification), use the following command:

make conduct

This will execute a silent, animated orchestration of the entire process, culminating in a single ✓ to confirm that the verification was successful.

---

Project Structure

• contracts/: The Solidity source code for the verifier and coordination contracts.
• scripts/: Deployment, interaction, and visualization scripts.
• src/: Python source code for the off-chain agents.
• docs/: Detailed project documentation.
• PHILOSOPHY.md: The core concepts and vision behind the project.
• ROADMAP.md: The strategic roadmap for future development.
• Dockerfile: A Dockerfile for creating a reproducible environment.
• LICENSE: The project's open-source license.