Three tools with basic descriptions and partial schemas. All tools have descriptions (10-50 chars range, below the 50-200 char LLM-optimized baseline), and input schemas are present but lack rigor. Parameters are documented but descriptions are minimal. Error handling returns raw exception strings rather than actionable recovery guidance. No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Output schemas are not formally documented, only inferred from code comments. Naming is clear (verb_noun pattern: run_report, run_realtime_report, get_metadata), but parameter descriptions lack format constraints, ranges, and examples of valid values.

• Expand tool descriptions to 100-150 characters. For run_report: 'Retrieve GA4 metrics and dimensions for a date range. Use for historical analysis, comparisons, and trend reports. Returns markdown table with rows and totals.'
• Add enum constraints to date_range parameter: document that string values must be one of ['last7days', 'last30days', 'today', 'yesterday'] and dict must have keys 'start_date' and 'end_date' in YYYY-MM-DD format.
• Document output schema explicitly in tool description: 'Returns a markdown-formatted table with columns for each dimension and metric, plus a totals row if available.'
• Replace generic error messages with actionable guidance: 'Invalid property_id: {property_id}. Verify the GA4 property ID in your environment or pass a valid property_id parameter.' Include retry classification (retryable vs user-fixable).
• Add tool annotations to schema: set readOnlyHint=true and idempotentHint=true for all three tools to signal to LLM that calls are safe to retry and do not modify state.

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

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