Three tools with reasonable descriptions (60-120 chars) but critical schema gaps. All tools lack explicit input parameter type definitions and descriptions in the visible code. The assess tool has a complex input object with nested properties (erc8004, oasf) but no schema validation visible. generateAgentURI and compare accept object parameters with no documented field structure. No output schemas documented. Error handling exists (Validation.error calls) but error messages are not visible in source. Tools follow verb_noun naming (assess, generateAgentURI, compare) but descriptions lack WHEN/WHY guidance for LLM selection.

• Add explicit JSON Schema for all input parameters. For assess: define erc8004 as {type: 'object', properties: {rpcNodes: {type: 'array', items: {type: 'string'}}}, ...}. For generateAgentURI: document assessmentResult structure (categories, entries fields and their types).
• Document output schemas for all tools. E.g., assess returns {status: 'string', messages: 'array', categories: {supportsMcp: 'boolean', ...}, entries: {...}, layers: {...}}.
• Enhance tool descriptions with selection criteria. E.g., 'Call assess() first to validate an MCP endpoint and gather metadata. Then call generateAgentURI() to create an EIP-8004 registration URI, or compare() to detect changes between two assessments.'
• Add parameter descriptions for complex objects. E.g., erc8004 param: 'ERC-8004 on-chain verification config. Object with rpcNodes (array of RPC endpoint URLs) and optional chainId (number). Required to enable on-chain agent registration checks.'
• Migrate from STDIO to HTTP+SSE or Streamable HTTP transport to enable remote client access and hosted deployment scenarios.