Two tools with basic schemas and minimal descriptions. generate_data_chart has a reasonable schema with enum constraint on chartType, but descriptions are generic (10-20 chars). generate_ai_chart has only a single string parameter with no format guidance. Neither tool description explains WHEN to use it, WHAT it returns, or error recovery paths. No output schemas documented. Parameter descriptions are trivial ('Chart data with labels and datasets' is not actionable). No error handling guidance. Schemas present but incomplete, missing required field declarations, no output structure, no pagination or limits.

• Expand tool descriptions to 100-200 characters. Example: 'generate_data_chart: Render a bar, line, pie, or doughnut chart from structured data. Accepts labels array and datasets with numeric values. Returns base64-encoded PNG image. Use when you have pre-computed data; use generate_ai_chart for natural language descriptions.'
• Document the output schema explicitly. Specify: does the tool return {image: string, format: 'base64'} or {url: string, width: number, height: number}? Include this in the tool description and in a separate 'returns' field if the MCP server supports it.
• Add format and constraint guidance to generate_ai_chart prompt parameter. Example: 'Natural language description of the chart (e.g., "bar chart showing Q1-Q4 sales by region"). Supported chart types: bar, line, pie, doughnut. Keep under 200 characters.'
• For generate_data_chart, specify the exact structure of the 'data' parameter. Example: '{labels: [string], datasets: [{label: string, data: [number]}]}'. Add min/max constraints on array sizes.
• Add error recovery guidance to both tool descriptions. Example: 'If chart generation fails, verify data format matches expected schema. For AI charts, try a simpler description or specify the chart type explicitly.'

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

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