Kalshi MCP has clear, well-structured tool definitions with proper naming (verb_noun pattern) and reasonable descriptions (avg 150 chars). All 4 tools have input schemas with types and descriptions. However, output schemas are not documented, LLMs cannot predict response structure for downstream chaining. The place_order tool lacks explicit error guidance for common failures (insufficient balance, invalid price range). Parameter descriptions are adequate but could be more prescriptive about constraints (e.g., price_cents range 1 - 99 is stated but not validated). No tool annotations (readOnlyHint/destructiveHint) despite clear read vs. write semantics.

• Document output schemas for all tools. Example: get_markets returns {markets: [{ticker, title, yesPrice, noPrice, volume, closeTime, ...}], total: number}. Include field types and whether pagination is supported.
• Add tool annotations: mark get_* tools with readOnlyHint=true; mark place_order with destructiveHint=true and idempotentHint=false (retrying with same params may double-execute).
• Enhance place_order description: 'Places a limit buy order. If dry_run=true, validates parameters and returns estimated execution details without executing. If dry_run=false, executes immediately and returns order_id. Errors: insufficient balance, invalid price (must be 1 - 99 cents), market closed.'
• Add input validation with actionable errors. If limit > 200, return 'limit must be 1 - 200, got 500' not a silent cap. If price_cents outside 1 - 99, return 'price_cents must be 1 - 99 cents, got 150'.
• Document get_positions response structure: does it return {positions: [...], total: number}? Include fields like ticker, side, count, avg_price, pnl. Clarify whether settled positions are included by default or only when settled=true.