Three tools with complete schemas and clear descriptions. Naming follows verb_noun convention (search_genes, get_protein, top_tissues). Descriptions are concise (100-180 chars) and explain purpose and prerequisites. All parameters have types and descriptions. However, output schemas are not documented, LLMs cannot predict response structure for planning. Error handling is minimal (generic error objects). No enum constraints on parameters. Parameter descriptions lack format/range guidance (e.g., 'limit' has no bounds). No idempotency hints or destructive operation warnings. Keyless design is good for accessibility but not mentioned in tool descriptions.

• Document output schemas for each tool. Example for search_genes: {count: number, genes: [{gene: string, ensembl: string, synonyms: string, description: string}]}. This enables LLMs to plan multi-step workflows.
• Add min/max bounds to numeric parameters. E.g., limit: {type: 'number', description: 'Max genes to return (1 - 100, default 15).'}
• Enhance error responses with recovery guidance. Instead of {error: 'protein not found'}, return {error: 'protein not found', ensembl_id: '...', suggestion: 'Use search_genes() to find the correct Ensembl ID.'}
• Remove example IDs from parameter descriptions. Replace 'An Ensembl gene id like "ENSG00000146648"' with 'An Ensembl gene id (format: ENSG followed by 11 digits).'
• Add tool annotations to the McpToolDefinition interface: readOnlyHint: true for all three tools. This signals to agents that these calls are safe to retry and do not modify state.