Etherscan just quietly shipped one of the most consequential infrastructure upgrades in recent memory, and it has nothing to do with a token launch. The blockchain explorer rolled out its “Build with AI” suite in late August 2026, giving AI agents and coding assistants like Claude and ChatGPT a direct pipeline to on-chain data across more than 60 EVM-compatible chains.
The centerpiece is a Model Context Protocol (MCP) server, essentially a standardized interface that lets AI models query blockchain data the same way a developer would, but without the duct-tape integration work that typically comes with wiring up APIs to language models.
What the MCP server actually does
The official endpoint lives at mcp.etherscan.io/mcp and exposes roughly 20 tools covering the core building blocks of blockchain interaction: balance queries, transaction lookups, token transfer histories, contract details, gas prices, and event logs. All accessible through a single API endpoint, authenticated with a standard Etherscan API key.
One standout feature is “Etherscan Flow,” a tool designed for tracing money movements across addresses. Transaction debugging tools are also part of the package, paired with machine-readable documentation that makes it easier for AI systems to understand what each endpoint returns and how to use it correctly. The entire suite operates in read-only mode and respects standard API quotas and rate limits.
