Etherscan has rolled out a Claude Connector that links its blockchain data platform to Claude, the AI model built by Anthropic. Users can query wallet and transaction data through natural language prompts. Etherscan launched the connector publicly on October 21, 2026.
The connector covers on-chain information across more than 60 Ethereum Virtual Machine (EVM)-compatible chains. EVM-compatible chains run smart contracts using the same execution environment as Ethereum, which is why one tool can read data from all of them.
The list of supported queries includes wallet balances, transaction histories, gas prices, event logs and contract verification.
Token transfers are covered across three common standards: ERC-20, ERC-721 and ERC-1155. ERC-20 covers fungible tokens like stablecoins. ERC-721 and ERC-1155 handle NFTs and mixed collections of digital items.
The integration requires neither an API key nor an Etherscan account.
Read-only by design
The connector is strictly read-only. It retrieves information. It does not sign transactions, move funds or interact with wallets in any active way.
Responses come with direct links back to Etherscan, so users can check the underlying records themselves rather than taking the AI’s word for it.
How we got here
The Claude Connector first landed in Claude’s marketplace directory in September 2026, weeks before the public launch.
In August 2026, Etherscan launched a Model Context Protocol (MCP) server as part of its broader “Build with AI” initiative. An MCP server acts as a bridge, giving AI assistants a structured way to request data from an outside service.
Etherscan still offers a keyless version of that MCP server. It includes 16 tools, with rate limits applied to keep usage in check. Developers who want more can unlock additional PRO features through authenticated access on paid API plans.
The resulting structure is tiered. The Claude Connector serves people who want answers without setup. The keyless MCP server serves tinkerers. Paid API access serves teams that need heavier, more advanced usage.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.