9 tools with clear verb-noun naming (get_, generate_, check_, convert_, simulate_). All have descriptions (avg ~120 chars, within baseline 34-392). Input schemas present with type definitions. However, descriptions lack LLM-optimized guidance on WHEN to use each tool vs alternatives; parameter descriptions are minimal (e.g., 'Hex color code' repeated 5 times); no output schemas documented; error handling returns generic dicts without recovery guidance; no tool annotations (readOnlyHint, etc.). Composition is sound (each tool does one thing), but descriptions don't explain dependencies or prerequisites. Baseline: 194 chars avg description, 72 chars avg param annotation, this server averages ~50 chars per param description.

• Add output schema to every tool definition. Document return type, field names, and types. E.g., tool_check_contrast returns {foreground: string, background: string, contrast_ratio: number, wcag_aa_normal: boolean, rating: 'AAA'|'AA'|'AA Large'|'Fail'}.
• Expand parameter descriptions to 50-100 chars with format constraints. E.g., 'Hex color code (6 digits, e.g., #ff0000 or ff0000; case-insensitive)' instead of 'Hex color code'.
• Add recovery guidance to error responses. E.g., if hex validation fails, return {error: 'Invalid hex: got "xyz", must be 6 hex digits (0-9, a-f). Example: #ff0000', suggestion: 'Try tool_convert_color() to convert from RGB or HSL.'}.
• Declare tool annotations in schema: all 9 tools should include readOnlyHint: true in their definition to signal safety to agents.
• Add WHEN-to-use guidance to tool_get_color_info and tool_get_color_info_local descriptions. E.g., 'Use local variant for fast offline color conversions (RGB, HSL, HSV, luminance). Use API variant for detailed color names and CMYK data from TheColorAPI.'