Three well-named, read-only tools with complete input schemas and clear descriptions. All tools follow verb_noun naming (search_*, get_*). Descriptions are concise (80 - 180 chars) and explain purpose clearly. Input schemas are properly typed with required fields marked. However, output schemas are not formally documented, LLMs must infer response structure from descriptions. Error handling is basic (generic error objects). No parameter constraints (enums, ranges) despite opportunities (e.g., fuzzy and marine_only could be documented more formally). No tool annotations (readOnlyHint, idempotentHint) despite all being read-only and idempotent.

• Document output schemas formally. For search_species, specify: { count: number, species: [{ aphia_id: number, name: string, authority: string, status: string, rank: string, kingdom: string, family: string, genus: string, valid_name: string, accepted_aphia_id: number }] }. Repeat for get_classification and get_common_names.
• Add tool annotations to all three tools: readOnlyHint: true and idempotentHint: true. This signals to agents that these calls are safe to retry without side effects.
• Add limit parameter to search_species (default 20, max 100) and document pagination. Return total count and offer cursor-based pagination for large result sets.
• Enhance error messages with recovery guidance. E.g., if get_classification returns 'not found', suggest 'Try search_species() to find the correct AphiaID first.'
• Formalize boolean parameter constraints. Use enum: [true, false] in schema or add explicit constraint text: 'fuzzy: boolean (true for LIKE substring matching, false for exact match; default false).'