Etherscan Transaction Debugger Skill Guides Agents Through EVM Analysis

Etherscan’s agent-skill repository includes a Transaction Debugger that helps AI agents explain EVM activity from transaction records while requiring them to disclose when execution evidence is incomplete.

An agent starts with a 32-byte transaction hash and can add a second hash to compare transactions. The workflow gathers transaction and receipt records, status information, contract and ABI details, token and label information, and links to explorer records. It then reconciles asset movements and permissions.

The skill offers five explanation levels: simple, standard, developer, support and security. It tells agents to mark steps inferred from the available evidence. Internal-transaction records should not be treated as a complete execution trace, and a call tree should be described as partial when a full trace is unavailable.

For failures, the instructions call for the narrowest root cause supported by the evidence. Agents should not identify an exact reverting internal frame without direct evidence. The skill also instructs agents not to sign or broadcast transactions.

Etherscan lists Flow for broader wallet and fund-flow investigations and Contract Review for reviews of verified deployed contracts.