Etherscan Launches AI On-Chain Data Toolkit
Etherscan has released the "Build with AI" toolkit, allowing developers and AI agents to access certified real-time on-chain data across more than 60 EVM-compatible chains via MCP, CLI, and Skills.
The toolkit connects to the same Etherscan API data layer as the etherscan.io website, supporting queries for on-chain native coin and token balances, regular transactions, internal transactions, transaction receipts, contract information, Gas prices, and event logs. The available endpoints and access permissions for different chains still depend on the user's API plan.
Etherscan MCP is aimed at agents that can call tools like Claude, ChatGPT, and Codex, enabling developers to request on-chain information in natural language through the official MCP endpoint. This service offers 20 tools that can cover all networks supported by Etherscan in a single connection, authenticated via API key.
Etherscan CLI can be used in terminals, scripts, and CI processes, generating JSON by default while also supporting table and CSV formats; its purpose is to allow developers to embed queries for addresses, transactions, contracts, and on-chain events into automated development and operational processes without requiring agents to scrape web pages or parse unstructured pages.
The Skills module provides installable professional task packages, including Orchestrator, Contract Review, Transaction Debugger, and Etherscan Flow. Flow can trace the flow of funds between addresses and write case files, the contract review module explains the behavior of verified contracts, the transaction debugger restores the execution content and failure reasons of transactions, while Orchestrator routes tasks to suitable tool interfaces.
Etherscan emphasizes that Skills require addresses, amounts, and transaction hashes to be obtained through real-time API calls, rather than being completed by the model; the toolkit is read-only and adheres to existing API usage quotas and rate limits. Developers can use the official llms.txt, llms-full.txt, or Docs MCP to provide contextual documentation to agents, reducing the risk of misuse of interface parameters, chain IDs, and data formats.
In terms of market mechanisms, on-chain data service providers are shifting their business, originally aimed at human web browsing and developer REST calls, to a structured data infrastructure for agents. Developers and security teams can delegate contract reviews, transaction troubleshooting, fund flow tracking, address monitoring, and preliminary risk control screening to AI workflows, with demand shifting towards Etherscan API plans, proxy development tools, wallet security services, compliance analysis, and security audit platforms; meanwhile, API limits, data latency, index integrity, and model misinterpretation of on-chain semantics will still restrict automated decision-making. Beneficiaries will be platforms that possess high-quality indexed data, developer distribution channels, and workflow integration capabilities, while services relying on manual web searches and screenshot-based on-chain analysis will face efficiency pressures.
Source: Public Information
ABAB AI Insight
Etherscan's core product in the past was the block explorer: it indexed raw blocks, transactions, logs, contract source code, and address activities into web pages and APIs that could be searched by humans. With the expansion of the multi-chain EVM ecosystem, this capability has gradually evolved from a "web tool for viewing transactions" to an underlying indexing layer used by applications, wallets, audit teams, and data products. The encapsulation of the same API into MCP, CLI, and Skills means that Etherscan is no longer just serving developers who directly call interfaces, but is attempting to occupy the default source of on-chain facts in AI coding tools and agent workflows. Its competitors are not only other explorers but also Alchemy, Infura, Covalent, The Graph, and various on-chain data warehouses.
In terms of capital pathways, MCP connects AI assistants to a paid API resource pool, CLI enters scripts, development environments, and continuous integration systems, while Skills solidify high-frequency work into reusable task templates. These three layers compete for use cases in chat-based queries, engineering automation, and complex case handling; users pay for API keys, call quotas, broader network coverage, and higher throughput, while Etherscan transforms data indexing, node access, standardization, and authentication capabilities into measurable services consumable by agents. Although free API keys can cover all supported networks, access to different endpoints and chains is still subject to plan limitations, indicating that the AI entry will not eliminate the commercial stratification of data infrastructure but may increase the demand for higher-tier quotas due to proxy high-frequency polling.
Historically, Bloomberg Terminal became the default information entry for financial practitioners through unified data formats, identifiers, and workflows, while GitHub became the record layer for software collaboration through code hosting, Pull Requests, and CI integration. Etherscan is attempting to replicate a similar position: not training a model that "understands crypto," but providing verifiable addresses, balances, transaction hashes, contract code, and event logs as the factual boundaries when AI executes tasks. Compared to allowing models to directly summarize block explorer pages, structured tool calls are easier to retain query conditions, output fields, and time points, suitable for forming auditable risk control and research processes; however, it cannot automatically verify external labels, project entity relationships, or attack attribution.
Essentially, this is a reconstruction of the industry chain: as AI agents enter on-chain operations, the scarcest resource is not the natural language ability to "explain blockchain," but a trustworthy, real-time, traceable, and unified interface for data entry. Models can be replaced, prompts can be copied, but a cross-chain indexing pipeline covering over 60 chains, stable API authentication, call quotas, contract verification records, and tool ecosystems require long-term maintenance. The mechanism of occurrence is that on-chain assets and smart contracts have public verifiability, but there are also issues of massive data volume, dispersed cross-chain formats, and complex transaction semantics; MCP and Skills shift this complexity from developers and models to data providers. Whoever controls this standardized entry will gain data distribution and service pricing power between wallets, audits, transactions, compliance, and AI agents.
ABAB News · Cognitive Laws
AI generates answers, but data determines whether the answers can be accountable
On-chain transparency is not readable; structuring makes it usable
Models can be replaced, but factual entry is the hardest to replace