Server has 19 tools with consistent naming (gsc_* prefix) and descriptions present for all. However, parameter descriptions are sparse or generic, and output schemas are not documented in the visible code. Most tools accept 3-6 parameters with type info but lack detailed constraints (enums, ranges, formats). Error handling is basic. Tool composition is sound (each does one thing), but descriptions could be more LLM-optimized per the 50-200 char baseline.

• Add JSON Schema output definitions for all tools. Document the structure of returned objects (e.g., gsc_search_analytics returns {rows: [{query, page, clicks, impressions, ctr, position}], rowCount, ...}). This enables LLMs to plan downstream calls and extract the right fields.
• Convert string parameter constraints to JSON Schema enums. For 'dataState', 'aggregationType', 'metric', 'type', and 'analysisType', add 'enum' arrays to the schema instead of relying on description text alone.
• Add min/max constraints to numeric parameters. E.g., rowLimit: {type: 'number', minimum: 1, maximum: 25000}, lookbackDays: {minimum: 2, maximum: 60}. This prevents LLMs from passing invalid values.
• Expand parameter descriptions with dependency hints and constraints. E.g., 'dimensions: Array of grouping dimensions (query, page, country, device, searchAppearance, date). Cannot be empty. Mutually exclusive with aggregationType=byPage.' This guides LLM reasoning.
• Add error handling guidance to tool descriptions. E.g., 'If siteUrl is not found, call gsc_list_sites() first to see available sites.' This prevents wasted calls and guides multi-step planning.

Inferred effective spec: 2025-06-18+.

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