10 tools with complete schemas and descriptions, but quality is inconsistent. All tools have basic descriptions (50-100 chars) and input schemas with types, but descriptions lack LLM-optimized guidance on WHEN to use each tool vs. similar ones. Parameter descriptions are minimal (e.g., 'Exchange name', 'Timeframe'). No output schemas documented. Error handling returns JSON stringified errors without recovery guidance. Tool names follow verb_noun pattern (scan, filter, analyze) but many are domain-specific jargon ('bollinger_scan', 'bbw_threshold') that may confuse LLMs unfamiliar with trading terminology. No tool annotations (readOnlyHint, destructiveHint). All tools are READ_ONLY per metadata, but this is not reflected in schema or description.

• Document output schemas for all tools. E.g., 'Returns array of {symbol: string, changePercent: number, price: number, volume: number}'. This enables LLMs to chain tools and extract relevant fields.
• Expand parameter descriptions to 50-100 chars with valid value guidance. E.g., 'pattern_type: Type of candle pattern to scan (bullish=green candles, bearish=red candles, neutral=mixed)' instead of just 'Pattern type'.
• Add WHEN-to-use guidance to tool descriptions. E.g., 'Use bollinger_scan to find low-volatility squeeze setups; use rating_filter to rank assets by technical strength. Call bollinger_scan first to identify candidates, then rating_filter to rank them.'
• Replace JSON.stringify(error) with structured error responses: {error: string, code: string, retryable: boolean, suggestion: string}. E.g., 'Invalid exchange: KUCOIN not found. Try: BINANCE, COINBASE, KRAKEN. Retryable: false.'
• Add tool annotations to schema. Mark all tools with readOnlyHint: true in the tool definition so LLMs know they are safe to call without confirmation.

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

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