Two tools with basic schemas and descriptions, but significant gaps in parameter documentation and output structure. Both tools have verb-noun naming (get-alerts, get-forecast) and non-empty descriptions (10-50 chars, below the 50-200 char LLM-optimized baseline). Input schemas are present with Zod validation and type constraints, but parameter descriptions are minimal. No output schema documentation. Error handling returns text responses but lacks actionable recovery guidance. No tool annotations (readOnlyHint, idempotentHint). Composition is sound, each tool does one thing, but responses are unstructured text rather than typed objects.

• Expand tool descriptions to 50-200 characters. Example: 'get-alerts' → 'Retrieve active weather alerts for a US state. Returns event type, severity, area, and headline. Useful for checking if dangerous weather is affecting a region.'
• Document output schema explicitly. Define a structured response type: {alerts: [{event: string, area: string, severity: string, headline: string}]} instead of free-text formatting.
• Add tool annotations in server.tool() calls. Use readOnlyHint: true for both tools (they do not modify state). Example: server.tool('get-alerts', ..., {readOnlyHint: true}).
• Enhance error messages with recovery guidance. Instead of 'Failed to retrieve alerts data', return 'Failed to retrieve alerts for state code CA. Verify the state code is valid (e.g., CA, NY, TX) and try again.'
• Add a limit parameter to get-alerts (default 10, max 50) to cap results and prevent context bloat. Document in description: 'Returns up to [limit] active alerts, newest first.'