Lido MCP Server + Vault Monitor Agent.

w's Team

claude-opus-4-6claude-codeMCP (Model Context Protocol) SDK@lidofinance/lido-ethereum-sdk@modelcontextprotocol/sdkviemzodtsup+3

Problem Statement

AI agents today cannot safely interact with DeFi protocols like Lido, and there is no way for agents to autonomously monitor vault positions for yield degradation, exploit signals, or governance risks. Without a standardized interface, agents risk executing unintended transactions; without monitoring, depositors miss critical changes. This project solves both: the MCP server provides a comprehensive, simulation-first toolkit for on-chain operations with full protocol awareness and human-in-the-loop confirmation, while the vault monitor agent watches Lido Earn vaults against user-defined bounds — flagging APR drops, share price anomalies, TVL shifts, and benchmark underperformance — making Lido position management accessible and safe for any MCP-compatible AI agent.

A production-grade MCP server and autonomous vault monitoring agent for the Lido staking protocol. The MCP server provides 60+ tools for staking ETH, managing positions, wrapping/unwrapping tokens, handling withdrawals, querying cross-chain balances, and participating in all 4 Lido governance systems (Aragon DAO, Snapshot, Easy Track, Dual Governance) — with dry-run simulation by default for safety. The vault monitor agent continuously watches Lido Earn vaults (ERC-4626), evaluating yield performance against benchmarks like raw stETH APR, detecting share price anomalies that could indicate exploits, tracking TVL shifts and protocol allocation changes, and firing alerts via email and Telegram when user-defined rules trigger. Rules use mathjs expressions for compound conditions (e.g. "apr < 3.0 and spread_vs_steth < 0") with templated messages. Together, they let any MCP-compatible AI agent stake, govern, and monitor Lido positions from a single conversation with no custom integration code.