Three well-named, read-only tools with clear descriptions and complete input schemas. Tool names follow verb_noun pattern (get_gene, search_genes, resolve_xref). Descriptions are detailed (150 - 250 chars) and explain WHEN to use each tool. All parameters have type definitions and descriptions. However, output schemas are not documented, LLMs cannot predict response structure. Error handling returns helpful messages but lacks recovery guidance. No tool annotations (readOnlyHint) despite all being read-only. Missing pagination guidance for search_genes despite returning ranked results.

• Document output schema for each tool. Example: get_gene returns {hgnc_id: string, symbol: string, name: string, locus_type: string, location: string, alias_symbols: array, ...}. Include field types and whether fields are always present.
• Add readOnlyHint and idempotentHint annotations to all three tools in the McpToolDefinition interface. This signals to agents that calls are safe to retry and have no side effects.
• For search_genes, clarify pagination: 'Returns up to limit results ranked by relevance score. If total > limit, call again with a higher limit or implement client-side pagination. No cursor support; use limit to control batch size.'
• Enhance error responses with structured recovery hints. Instead of {error: 'no approved gene...'}, return {error: 'no approved gene...', suggestion: 'Try search_genes(query="BRCA") to find similar symbols', retryable: false}.
• Add examples to parameter descriptions showing valid inputs: symbol: 'Official HGNC symbol (e.g., BRCA1, TP53, EGFR)' and id: 'External ID (e.g., entrez "672", ensembl "ENSG00000012048")' to guide LLM input generation.