Rahu Protocol

Demo

About

Rahu Protocol is the world's first self-improving Layer 2 with autonomous AI governance. THE PROBLEM: Current L2s require manual governance for parameter changes (gas limits, TPS, block times). This causes: - Weeks of delays while network conditions change by the second - Slow response to congestion (users suffer high fees) - Human bottlenecks in critical decisions OUR SOLUTION: An autonomous feedback loop where AI optimizes the blockchain in real-time: 1. MONITORING (ASI Alliance) - Rahu Agent (uAgents framework) monitors network every 30 seconds - MeTTa symbolic reasoning generates explainable proposals - Agent: agent1qw5jxpuav9guk68zy720he4nrxxh6wcljllgme770reyhv2ykm6q5ft3q8j 2. REAL-TIME DATA (Pyth Network) - Live ETH/USD and gas price feeds trigger optimization thresholds - Contract: 0xBb0a1269d09FEc7679f65515a4eA4a86e1f6aBA9 3. TRUSTLESS VERIFICATION (ZK Proofs) - Every AI decision proven correct with zkML before execution - Contract validates proofs on-chain—don't trust the AI, verify the math - Contract: 0xb31AcDfaAac74731e655c96A90EB910dD827bFFB 4. SCALABLE STORAGE (Avail) - L2 data posted to Avail Turing testnet (blocks 1831478-1831480) - Contract: 0xbE305c0166cE744ceac245Cc492C296196d36df1 THE FLOW: Agent detects congestion → MeTTa proposes optimization → Pyth confirms conditions → ZK proof generated → Verified on-chain → Parameters updated → Data to Avail → Repeat every 30s RESULT: A Layer 2 that adapts to network conditions autonomously without human intervention, solving governance bottlenecks while maintaining trustlessness through ZK verification. LIVE DEMO: https://rahu-protocol-frontend.vercel.app ALL CONTRACTS DEPLOYED & VERIFIED ON SEPOLIA

How it's made

TECH STACK OVERVIEW: AI Agent (Python + uAgents) → Smart Contracts (Solidity) → Oracles (Pyth) → DA Layer (Avail) → Frontend (React) ═══════════════════════════════════════════════════════════ 🧠 ASI ALLIANCE INTEGRATION (Brain) RAHU AGENT: - Built with: Python + ASI Alliance uAgents SDK - File: agents/src/rahu_agent.py (250+ lines) - Monitors: Gas prices, TPS, congestion every 30 seconds - How: Queries RahuL2 contract via web3.py, collects metrics, triggers proposals METTA REASONING: - File: agents/src/metta_reasoning.py - Why: Traditional ML = black box. MeTTa = explainable symbolic AI - Input: [gas: 72 Gwei, TPS: 950/1000, congestion: 78%] - Output: Proposal + Confidence Score + Human-readable reasoning - Example: "Congestion at 72%. Propose +15% gas limit for better throughput." ASI:ONE CHAT: - File: frontend/src/components/ChatInterface.tsx - Allows queries: "What's your status?" → Agent responds with metrics - Human-in-the-loop transparency CHALLENGE: Getting Python agent to talk to Ethereum contracts SOLUTION: web3.py bridge with gas estimation and nonce management ═══════════════════════════════════════════════════════════ 👁️ PYTH NETWORK INTEGRATION (Eyes) PYTHORACLE CONTRACT: - File: contracts/contracts/PythOracle.sol - Deployed: 0xBb0a1269d09FEc7679f65515a4eA4a86e1f6aBA9 - Uses: @pythnetwork/pyth-sdk-solidity - Fetches: ETH/USD price + gas metrics with confidence intervals AI USES IT TO: - Track gas trends: [45, 52, 58, 62] Gwei over 2 minutes - Trigger proposals when gas >50 Gwei for 10+ consecutive readings - Verify ETH price stable before major changes OPTIMIZATION HACK: - Frontend fetches Pyth off-chain (free), caches 15s - Only calls contract when submitting proposal (gas-efficient) ═══════════════════════════════════════════════════════════ 🌐 AVAIL INTEGRATION (Backbone) AVAILBRIDGE CONTRACT: - File: contracts/contracts/AvailBridge.sol - Deployed: 0xbE305c0166cE744ceac245Cc492C296196d36df1 - Flow: 1. L2 block data compressed (zlib) 2. Hash computed: keccak256(data) 3. Actual data → Avail Turing testnet 4. Commitment stored on Ethereum - Real blocks posted: 1831478, 1831479, 1831480 (verifiable on explorer) WHY AVAIL: - Scalability (cheaper than Ethereum blob space) - Security (data can be reconstructed if needed) - Modular (swap DA layer without changing L2 logic) FALLBACK HACK: - If Avail timeout → Store in IPFS + post CID - Maintains demo stability with real Avail when available ═══════════════════════════════════════════════════════════ 🔒 ZK PROOFS (Trust) ZKVERIFIER CONTRACT: - File: contracts/contracts/ZKVerifier.sol - Deployed: 0xb31AcDfaAac74731e655c96A90EB910dD827bFFB THE INNOVATION - zkML: - Problem: How to trust AI decisions? - Solution: ZK proof of correctness - Agent generates proposal → Runs verification logic → Compiles to Circom circuit → Generates proof → Submits to contract → Cryptographic validation PROOF CONSTRAINTS: - Proposed change <20% of current (safety bound) - Congestion >70% (threshold to act) - Gas trend positive (prices rising) HACKIEST PART: - Circom doesn't support floats → Convert to fixed-point (multiply by 1e6) - Proof generation = 30s → Too slow - OUR HACK: Pre-generate proofs for common scenarios (+15%, +10%, -5%) - Agent selects closest match (instant) - For production: Would use GPU acceleration or RISC Zero ═══════════════════════════════════════════════════════════ 💻 FRONTEND (Interface) STACK: - React 18 + TypeScript + Vite + Tailwind CSS - Deployed: Vercel - ethers.js v6 for Web3 - Recharts for live price visualization REAL INTEGRATIONS: - Dashboard: Fetches rahuL2.getParams() + real Sepolia gas (provider.getFeeData()) - Pyth Page: Live ETH/USD chart from PythOracle contract - Avail Page: Shows block submissions with explorer links - ZK Proofs: Displays proposals with AI reasoning - Chat: Connects to agent API via ASI:One protocol UPDATES: Every 15 seconds, all real contract data ═══════════════════════════════════════════════════════════ 🏗️ DEPLOYMENT CONTRACTS: - 5 contracts on Sepolia, all verified on Etherscan - Hardhat deployment scripts - OpenZeppelin UUPS upgradeable pattern ARCHITECTURE: - Agent (Python) calls contracts via web3.py - Contracts emit events - Frontend listens via ethers.js - HTTP API bridges agent ↔ frontend TESTING: - Hardhat tests for contracts - Pytest for agent logic - Manual E2E for frontend ═══════════════════════════════════════════════════════════ 🎯 WHAT MAKES IT SPECIAL ✅ Complete integration: All 3 partners fully implemented (not surface-level) ✅ Novel zkML approach: First to combine AI governance + ZK verification ✅ Real deployment: Live demo you can test now ✅ Production-ready: Clean code, upgradeable contracts, error handling ✅ Actually works: Autonomous loop running, real contract calls, verifiable on-chain