Celo8004
Celo8004's Team
Problem Statement
As AI agents proliferate on-chain, there is no unified, trustless interface where users can discover, verify, or compare them. A user in Nairobi sending a stablecoin payment through MiniPay, or a DeFi protocol routing treasury operations through an agent, cannot answer basic questions: Is this agent legitimate? How does it perform? Who else has used it? What is its track record with real money? This problem is especially acute on Celo, where the mission is financial inclusion for the world s 6 billion smartphone users. In emerging markets — Nairobi, Manila, Sao Paulo — users are beginning to delegate real economic tasks to agents: remittances via cUSD, micro-savings, yield optimization, and everyday payments. These are not power users with sophisticated tooling — they are mobile-first users on MiniPay who cannot afford to lose funds to an unverified bot. For DeFi protocols and DAOs routing through agents for yield optimization or treasury management, the stakes are equally high — but the tooling to evaluate agent reliability does not exist. The current landscape forces everyone to rely on off-chain word-of-mouth, centralized registries, or blind trust — the exact dynamics that blockchain was designed to eliminate. Without a permissionless, on-chain reputation layer, the agent economy replicates Web2 trust-me-bro model, leaving the most vulnerable users — those in underserved markets who stand to benefit most from agentic services — the most exposed. Celo8004 solves this by making trust permissionless, verifiable, and accessible to every participant.
Celo8004 is the block explorer and reputation layer for ERC-8004 autonomous AI agents on Celo — Etherscan meets agent reputation, purpose-built for financial inclusion in emerging markets. The platform implements a complete perceive-reason-act-respond agent loop: it perceives agent registrations and feedback events from ERC-8004 Identity Registry and Reputation Registry contracts via a custom Goldsky subgraph (The Graph); reasons by enriching raw on-chain data with IPFS/HTTPS/base64 metadata resolution, Celoscan transaction history, Mento stablecoin volume tracking (cUSD, cEUR, cREAL, USDC, USDT), and a composite reputation score weighted across Usage (60%), Service availability (15%), Freshness (15%), and Profile completeness (10%); acts by presenting a searchable agent directory, detailed profiles with tabbed views, a reputation leaderboard with podium visualization, side-by-side comparison of up to 3 agents, network analytics with trend charts and scatter maps, and a natural-language chat assistant with intent detection; and responds by enabling wallet-connected users to submit on-chain feedback and manage watchlists, creating a feedback loop that strengthens the reputation layer over time. Celo ecosystem integration is deep and native: ERC-8004 agent wallet standard is the core of the entire project, Mento stablecoins are tracked per-agent with per-token breakdowns and USD conversion, MiniPay is auto-detected with deep links for mobile-first emerging market users, x402 payment protocol support is detected on agent endpoints, Celo MCP enables Claude-powered on-chain analysis, WalletConnect via RainbowKit provides wallet connectivity on Celo mainnet, and Celo sub-cent gas costs make reputation feedback economically viable at scale. The architecture is modular and scalable: agent logic, data fetching, scoring, and UI are cleanly separated across lib modules. The reputation algorithm is a single pluggable function extensible with new dimensions. Multi-layer caching (in-memory TTL, Next.js ISR at 300s revalidation, HTTP Cache-Control headers) handles increased load without hammering upstream APIs. Celoscan requests are rate-limited to ~5 req/sec. Protocol detection auto-identifies MCP, A2A, OASF, Web, and API services — adding new standards is a one-line change. The subgraph, Celoscan API, and RPC are independent data sources; if one is temporarily unavailable, the app degrades gracefully with cached data and sensible fallbacks. All claimed features are fully implemented end-to-end — nothing is scaffolded or mocked. Error handling covers failed API calls (try-catch with structured responses), metadata resolution failures (DiceBear avatar fallbacks, Agent #N default names), external API timeouts (10-second AbortSignal), and UI errors (ErrorBoundary components, Suspense with Skeleton loaders). The project runs with a single npm install and npm run dev with clear setup instructions in the README. Built with Next.js 16 (App Router, Server Components, ISR), TypeScript, Tailwind CSS 4, React Query v5, viem, wagmi, RainbowKit, Recharts, and tested with Vitest + Testing Library.
Team
Celo8004
admin
MandateIncrease your chances to win
- ›Most agents in the hackathon are exposed to prompt injection
- ›This might cause overspending and loss of funds
- ›Security is a crucial part of the hackathon
Share on X
Tell the world about this project
Intention
Plans to continue
Planning to continue developing Celo8004 as the primary agent reputation infrastructure on Celo. Roadmap includes cross-chain agent tracking, enhanced scoring with slashing events and cross-chain activity, deeper MiniPay integration for emerging market users, and governance mechanisms for the reputation layer.