Three tools with clear verb-noun naming (list_icons, search_icons, get_platform_usage). All have descriptions and input schemas visible in src/index.ts. However, descriptions are brief (10-60 chars), lacking context on WHEN to use each tool or what downstream actions they enable. Parameters lack descriptions in most cases: list_icons has no parameters; search_icons.query lacks guidance on format/examples; get_platform_usage.platform has an enum but no description of what 'usage' means. No output schemas documented, LLMs cannot predict response structure for chaining. Error handling is minimal (McpError for unknown tools only). No pagination, no result limits stated. Resources are defined but tool-resource relationships are unclear.

• Document output schemas for all three tools. Example: list_icons returns {icons: [{name: string, tags: string[], category: string}], total: number}. This enables LLMs to chain tools and extract relevant fields.
• Expand tool descriptions to 100-200 chars. Example: 'Search for Hugeicons by name or tags. Returns matching icons with metadata. Use this to find specific icons; call list_icons first to explore all available icons.' This clarifies intent and relationships.
• Add descriptions to all parameters. Example: search_icons.query: 'Comma-separated icon names or tags (case-insensitive). E.g., "home, notification, settings". Returns up to 50 matches.'
• Add pagination to search_icons and list_icons. Accept limit (1-100, default 20) and offset parameters. Return total count and next_offset in response. Document in tool description: 'Results are paginated; default limit is 20.'
• Enhance error handling. Validate platform enum in get_platform_usage and return 'Invalid platform: got "xyz", must be one of: react, vue, angular, svelte, react-native, flutter.' For search_icons, handle empty results with 'No icons found for query "xyz". Try broader terms or call list_icons to explore all icons.'