Three well-named, read-only tools with clear descriptions and complete input schemas. Naming follows verb_noun pattern (search_datasets, query_layer, layer_info). Descriptions are substantive (150 - 200 chars) and explain what each tool does and when to use it. All parameters have type definitions and descriptions. However, output schemas are not documented, LLMs cannot predict response structure. Parameter constraints (e.g., limit 1 - 50, 1 - 2000) are stated in descriptions but not enforced via JSON Schema enums/min/max. No error handling guidance or recovery paths. Missing required field declarations in some schemas. SSRF guard in layerUrl() is a security plus.

• Add outputSchema to each tool definition documenting response structure. E.g., search_datasets returns {total: number | null, count: number, datasets: [{id, name, summary, org, org_id, record_count, type, url}]}.
• Enforce parameter constraints via JSON Schema: add 'minimum', 'maximum', 'enum' fields. E.g., limit: {type: 'number', minimum: 1, maximum: 50, default: 20}.
• Wrap error throws in structured responses with actionable guidance. E.g., {error: 'Invalid URL', message: 'Must be https and end in /FeatureServer/<n> or /MapServer/<n>', suggestion: 'Use search_datasets to find valid layer URLs.'}.
• Document query_layer pagination: clarify that limit caps results at 2000 and offset enables pagination. Suggest returning a next_offset or hasMore flag for large result sets.
• Add explicit timeout (e.g., 30s) to fetch calls and return timeout errors with retry guidance. Prevent agent hangs on slow Esri services.