Server has 4 tools with complete input schemas and basic descriptions in Japanese. However, descriptions are minimal (10-30 chars), lack actionable context for LLM selection, and parameter descriptions are sparse. No output schemas documented. Error handling is present but generic. Tool naming follows verb_noun pattern but lacks clarity on distinctions (e.g., analyze_property vs portfolio_analysis both perform analysis). No tool annotations (readOnlyHint/destructiveHint) despite clear risk profiles. register_property accepts untyped 'property_data' object, violating schema specificity. Missing parameter constraints (enums, ranges) for numeric inputs like loan_ratio, interest_rate.

• Expand tool descriptions to 100-200 chars. Example: 'analyze_property: Calculate ROI, cash flow, and profitability metrics for a single property. Use this to evaluate one investment opportunity. Returns annual_return, cash_flow_year1, roi_percent, and property_id for chaining to compare_properties.'
• Add parameter descriptions for all numeric inputs. Example: 'loan_ratio: Financing ratio as decimal (0.0=all cash, 1.0=100% financed). Typical range 0.6-0.9. Default 0.8.'
• Document output schema for each tool. Example: analyze_property returns {property_id, annual_gross_income, annual_expenses, annual_net_income, roi_percent, cash_flow_year1, payback_years}.
• Add enum constraints to loan_period (e.g., [15, 20, 25, 30]) and investor_tax_bracket (e.g., [0.1, 0.2, 0.3, 0.4]) to guide LLM input.
• Implement tool annotations in server.py: add readOnlyHint=true to analyze_property, compare_properties, portfolio_analysis; add destructiveHint=true to register_property. This requires updating Tool objects to include these fields.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry