Server defines 5 tools with complete input schemas and descriptions. All tools follow verb_noun naming (get_*, list_). Descriptions are adequate (100-200 chars) and explain purpose clearly. However, output schemas are completely undocumented, LLMs cannot predict response structure. Parameter descriptions lack format constraints (e.g., address format, chain ID ranges). No error handling guidance. Missing tool annotations (readOnlyHint). Composition is sound: each tool has one responsibility, and chain parameters enable multi-chain support.

• Document output schemas for all tools. Example for get_balance: 'Returns {address: string, balance: string (wei), chain_id: number, symbol: string, decimals: number}'. This enables LLMs to extract and chain results.
• Add format constraints to parameter descriptions. E.g., 'address: 0x-prefixed hex string, exactly 42 characters' and 'chain: numeric ID (1-534352) or slug from {ethereum, polygon, bsc, ...}'.
• Implement error handling with recovery guidance. Catch common failures (invalid address, unverified contract, rate limit) and return actionable messages like 'Contract not verified on this chain. Try get_contract_source on a different chain.' or 'Rate limited. Retry after 60 seconds.'
• Add readOnlyHint annotation to all tools in the tool definition. This signals to agents that these calls are safe to retry and have no side effects.
• Enhance list_transactions description to clarify pagination: 'Returns up to 10,000 results per page. Use page (1-indexed) and offset (results per page, 1-10000) to iterate. Defaults: page=1, offset=25.'