Server has 4 tools with basic schemas and descriptions. All tools are READ_ONLY with proper annotations. Naming follows verb_noun pattern (umami_list, umami_get, umami_whoami). Descriptions are present but minimal (10-50 chars), below the 50-200 char LLM-optimized baseline. Input schemas exist but lack detail: pageSize and period parameters have no enums, ranges, or format constraints. Output schemas are completely undocumented, LLMs cannot predict response structure. Error handling returns generic failure messages without recovery guidance. No parameter validation rules documented (e.g., what values are valid for 'period'?). Tool composition is sound (single responsibility), but descriptions need expansion to guide LLM selection.

• Document output schemas for all tools. Example for umami_list_websites: 'Returns {websites: [{id: string, name: string, domain: string, createdAt: ISO8601}], total: number, page: number}'.
• Add enums to period parameters: period must be one of '24h', '7d', '30d', '90d'. Replace free-form strings with constrained choices.
• Expand tool descriptions to 80-150 chars. Example: 'umami_get_stats: Retrieve visitor count, page views, bounce rate, and session duration for a website over a specified time period. Use this to answer questions about traffic trends.'
• Enhance error messages with recovery guidance. Instead of 'umami_get_stats failed: 404', return 'Website not found. Verify websiteId is correct by calling umami_list_websites first.'
• Add numeric bounds to pageSize: 'pageSize (1-100, default 20): Number of websites per page. Larger values may timeout.'

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

• Relies on Dynamic Client Registration (deprecated) - use Client ID Metadata Documents (CIMD)