Server has 4 tools with significant quality gaps. Tool naming is inconsistent (print_name vs get_ncm_info vs get_ncm_info_tool vs get_ncm_attributes_tool), three tools operate on NCM data but lack clear differentiation. Descriptions are present but generic; most lack actionable context for LLM selection. Schemas are partially defined: print_name has empty input {}, get_ncm_info and get_ncm_info_tool have proper string params, but get_ncm_attributes_tool uses SkipValidation[any] which bypasses validation entirely. No output schemas documented. Error handling is minimal, get_ncm_attributes_tool returns dict with 'error' key but no recovery guidance. No pagination, no result limits, no idempotency markers.

• Consolidate get_ncm_info and get_ncm_info_tool into a single tool with a clear name (e.g., get_ncm_details) and document the return type (dict with ncm, description, info fields).
• Replace SkipValidation[any] in get_ncm_attributes_tool with explicit type: string, and add parameter description: 'NCM code (8-digit numeric string, e.g. 12345678)'.
• Add output schema documentation to all tools. Example for get_ncm_attributes_tool: 'Returns HTML table string with columns: attribute_name, attribute_value, description. Returns {"error": "<message>"} if NCM not found.'
• Enhance error messages with recovery guidance. Example: 'NCM 99999999 not found. Call search_ncm() with a partial code to find valid NCM codes.'
• Add pagination support to get_ncm_attributes_tool if result sets can be large. Include limit (default 20, max 100) and offset parameters, and return total_count in response.