The MCP Chart Server exposes 27 chart generation tools with significant quality gaps. While some tools (generate_area_chart, generate_bar_chart, generate_column_chart, generate_dual_axes_chart, generate_funnel_chart, generate_histogram_chart, generate_mind_map, generate_network_graph) have partial schemas visible in callTool.ts, the majority (15 of 27 tools) lack documented input schemas entirely. Descriptions are present but generic and under-specified ('Generate a X chart visualization'). No output schemas are documented. Parameter descriptions exist for tools with schemas but are minimal and lack validation guidance. The rubric baseline shows 100% of A+ tools have documented return types and full parameter documentation; this server achieves ~30% schema coverage and ~40% parameter clarity. Tool naming follows verb_noun convention (positive), but compositions lack error guidance and recovery patterns.

• Add complete JSON Schema input definitions for all 15 undocumented tools (generate_line_chart, generate_pie_chart, etc.). Inspect the corresponding chart type implementations in src/charts/ and export schema from each.
• Expand all 27 tool descriptions from 35 chars to 100-200 chars. Example: 'generate_area_chart: Generates a stacked or non-stacked area chart suitable for time-series data or composition over time. Use this when you need to visualize cumulative trends or compare multiple categories over time. Returns an SVG visualization or remote image URL depending on renderer setting.'
• Document output schema for all tools. Add to tool descriptions: 'Returns {content: [{type: "text", text: <image-url-or-svg-string>}], _meta: {renderer: "remote"|"recharts", spec: {type, ...args}}}'. Clarify which renderer produces which output format (remote → imageUrl, recharts → svg, etc.).
• Add parameter constraints to documented schemas. Example for 'data' param: 'Array of objects with required fields: category (string), value (number). Optional: group (string for stacked variants). Maximum 1000 items to avoid rendering performance issues.'
• Implement error recovery guidance. Modify callTool error responses to include suggestions: 'Invalid parameters: category field missing. Ensure data is [{category: "Jan", value: 100}, ...]. See generate_bar_chart docs for examples.'
• Add a discovery tool: 'list_chart_types' returning {types: [{name: "area", description: "...", best_for: "...", required_fields: [...]}]}. This enables LLMs to recommend appropriate chart types before rendering.

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

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