Three tools with complete input schemas and descriptions. Tool naming follows verb_noun pattern (dealdesk_run_*, dealdesk_fund_*, dealdesk_price_*). Descriptions are substantive (100-200 chars) and explain the financial domain clearly. All parameters have types and descriptions with constraints. However, output schemas are documented only in docstrings, not formally in the tool registration. Error handling is minimal, no recovery guidance or categorization. Missing parameter validation examples and no guidance on when to use each tool vs. alternatives.

• Add formal response_schema to each tool definition. Document the JSON structure (e.g., for dealdesk_run_waterfall: {structure: string, total_proceeds: float, lp_distribution: float, gp_distribution: float, gp_carry: float, lp_moic: float, effective_carry_pct: float, tiers: [{name, lp, gp, total}]}).
• Wrap tool logic in try-except blocks that return structured error objects: {error: string, category: 'retryable'|'user_fixable'|'fatal', suggestion: string}. E.g., 'Invalid structure: got "british", must be "european" or "american". Try again with a valid structure.'
• Enhance parameter descriptions to include constraint ranges inline. E.g., 'GP co-investment as fraction between 0 and 1 exclusive (0.02 = 2%). Must be >= 0 and < 1.'
• Add a 'When to use' section to each tool description. E.g., 'Use dealdesk_run_waterfall to compute LP/GP splits for a single exit event. Use dealdesk_fund_metrics to analyze multi-period cash flows and IRR/MOIC over the fund lifetime.'
• Document max array length for cashflows parameter: 'cashflows: array of 2 - 100 equal-period cash flows (negative=contribution, positive=distribution).' Add validation to reject arrays > 100 items.