Aave Launches MCP Server for AI Agents to Access V3 and V4 Protocol Data

• Aave Labs has launched a Model Context Protocol server providing standardized, non-custodial access to live lending data across Aave V3 and V4.
• The infrastructure enables AI agents to query positions, simulate risk scenarios, and prepare unsigned transactions without holding private keys.
• Governance is currently voting to delegate limited V4 risk controls to Risk Stewards and evaluate emergency freeze powers for active exploits.
• Aave is simultaneously executing a strategic rationalization by closing six underperforming V3 markets while launching V4 on Avalanche for tokenized real-world assets.

Aave Labs has officially launched a Model Context Protocol (MCP) server, establishing a single, standardized endpoint for AI applications to access live lending data from its V3 and V4 infrastructure. This deployment replaces previous reliance on static datasets and fragmented third-party wrappers, effectively creating a universal translator between large language models and Aave’s on-chain systems. The server is accessible at mcp.aave.com and connects to Aave V3 deployments across 21 blockchains, as well as V4 on Ethereum and Avalanche. It offers approximately 40 tools that allow users and AI agents to retrieve market data, check wallet positions, examine health factors, and simulate potential actions before committing capital. This move positions Aave as a foundational data layer for AI-powered decentralized finance applications.

The architecture of the MCP server is strictly non-custodial, addressing a critical trust and security challenge in the agentic DeFi sector. The server can assemble and prepare unsigned transactions but never holds private keys or signs on behalf of the user. This design ensures that while an AI agent can manage risk and prepare execution steps, human approval or a separate signing module is still required for final transaction execution. Users can interact with the system through natural language queries, asking assistants to check wallet health factors, compare stablecoin supply rates, or simulate actions to preview resulting health factors before execution. Crucially, no actions execute automatically; every prepared transaction remains unsigned, requiring the user's wallet to complete the signing process. This ensures assistants do not move funds independently, maintaining the core security principle of self-custody in DeFi.

Chainlink provides the underlying real-time data feeds that keep these AI agents informed, ensuring the integrity of the information processed. While Aave built the connection infrastructure, Chainlink’s data layer supports the information layer, allowing AI agents to accurately report live rates and health factors to users. This development marks a significant shift in user interaction with DeFi, moving from direct website interfaces to AI-mediated access. By lowering the technical barrier through natural language interactions, the integration may encourage broader adoption of DeFi protocols among users who are less familiar with complex smart contract interfaces. The integration supports the growing DeFi AI race by providing comprehensive tools for market data, position management, and risk simulation.

How Will Emergency Powers Change Aave Governance?

Aave DAO voters are currently deciding whether to delegate limited V4 risk controls and no-delay emergency roles to Risk Stewards. The proposal assigns constrained changes that can be made without taking every update through a full governance vote. However, approval would not activate the system by itself; Aave Labs stated that corresponding payloads would still need to be executed through the V4 Security Council. The Risk Steward contracts are undergoing a Certora audit nearing finalization. The proposal's central tension is between authority assigned now and functionality available later. Each Risk Steward would receive Hub and Spoke risk-management roles plus emergency roles with no execution delay.

The current release calls none of the emergency selectors, leaving those permissions inert until a future release adds support. This design allows later versions to respond to emergencies without waiting through another governance cycle. The wider permission redesign splits V4 configurator controls into five granular categories: two flag-control roles, a listing role, an emergency role, and a risk-management role. Emergency permissions are limited to one-way safety actions, such as deactivating assets or pausing reserves, which cannot be reversed by the stewards. Routine parameter updates would operate under different controls with minimum cooldowns of 36 to 72 hours, capping how far each update may move parameters like interest rates or collateral factors. Aave governance is also considering an emergency Guardian powers proposal that would allow vulnerable lending pools to be frozen quickly during active security threats without requiring immediate public write-ups. This approach addresses the tension between DeFi transparency and security, arguing that immediate disclosure during an exploit can hand attackers a roadmap, while delayed disclosure protects users while the threat is contained.

What Strategic Shifts Are Driving Aave’s V4 Expansion?

Aave is undergoing a significant structural adjustment to optimize its protocol efficiency and risk profile. Following recommendations from DeFi risk management service LlamaRisk, Aave is evaluating the closure of six V3 blockchain markets on Sonic, Scroll, zkSync, Metis, Soneium, and Aptos. This governance proposal aims to retire 50 asset markets that have suffered sharp deposit declines, with some chains seeing over a 90% reduction. The move is designed to concentrate resources on high-liquidity markets, thereby reducing systemic risk and improving capital efficiency. Concurrently, LlamaRisk has proposed reserve-cap and interest-rate changes for Aave V3, recommending raised limits for high-demand assets like USDC and GHO while reducing limits for underused assets.

On July 15, 2026, Aave launched its V4 protocol on Avalanche, marking the first time the platform has expanded beyond Ethereum. This deployment introduces new lending infrastructure specifically designed to support tokenized real-world asset markets, positioning Aave at the intersection of DeFi and traditional finance. The V4 upgrade introduces a hub-and-spoke architecture that replaces isolated pools with a central liquidity hub per blockchain. User-facing 'Spokes,' such as Core or institutional Prime markets, tap into this shared pool, deepening available liquidity while keeping risk parameters like loan-to-value ratios separate and governable. Additionally, Aave introduced Stable Vaults to bridge DeFi with mainstream fintech. These vaults allow fintech companies, such as wallets and payment apps, to offer yield on customer stablecoin deposits by integrating Aave’s DeFi infrastructure. These developments reflect a dual strategy: retrenching from underperforming, low-revenue chains to strengthen core market security, while simultaneously expanding into high-growth areas like RWA tokenization and fintech partnerships.