1. Uniswap v4 Hook Integration: Integrating dynamic fee hooks into both swap and liquidity flows required deep understanding of Uniswap v4’s architecture. We overcame this by carefully reviewing the v4 router documentation, encoding pool keys manually, and simulating transactions on Base Sepolia.

2. Real Wallet Connectivity: Ensuring seamless wallet connection across OnchainKit, AgentKit, and Nebula was difficult. Wallet connection state needed to persist across chat, swap, and liquidity modals. We resolved this by standardizing wallet access across components using Wagmi context and OnchainKit’s smart wallet utilities.

3. Executing Transactions via AI Chat: Connecting Nebula to execute transactions — not just simulate them — required rewriting the agent execution logic and integrating it with our Uniswap v4 hook-based backend.

4. Token Metadata and Icon Issues: Displaying live token icons across swaps, liquidity, and portfolios was unreliable at first due to inconsistent API responses from Moralis. We solved this by implementing fallback token maps, batching API requests, and eventually introducing Thirdweb’s token API for stability.

5. React Render Loops & Modal Bugs: During the liquidity modal buildout, we hit several “too many re-renders” errors due to React state updates in render cycles. We addressed this with precise useEffect hooks and improved conditional rendering logic.

6. **Onchainkit & Agentkit: ** During development, we faced significant challenges integrating OnchainKit and AgentKit. The wallet system initially caused issues where transactions were executed using the wrong address, breaking user trust. We resolved this by restructuring wallet state management and tying all actions explicitly to the connected wallet. AgentKit presented another hurdle—its intent-based architecture conflicted with wallet execution flow, so we redesigned the experience so agents prepare transactions, but users must confirm execution via their wallet. We also implemented dynamic syncing between agent memory and wallet changes to prevent context mismatches. These fixes ensured secure, reliable, and intuitive user interactions.

Despite the complexity, each obstacle helped us refine the architecture into a scalable, modular DeFi+AI stack.