Three tools with reasonable naming (verb_noun pattern) and basic schemas, but significant gaps in parameter descriptions and output documentation. Tool descriptions are present but generic (10-50 chars in some cases). Parameters lack type constraints (enums, ranges). No error handling guidance. Schema completeness varies: find_protein_structures has good input schema; get_protein_data_tool and download_protein_structure_tool lack documented output schemas. Missing descriptions for several parameters (e.g., 'data_types' array lacks item type details). No mention of idempotency, retry safety, or side effects for the WRITE tool.

• Document output schemas for all tools. Example: 'Returns {pdb_id: string, title: string, exists: boolean, validation_result: string}' for find_protein_structures_tool.
• Convert 'data_types' parameter to an enum: ["basic", "sequence", "structure", "all"] with descriptions for each option.
• Add error handling guidance: 'If PDB ID not found, returns error with available search suggestions. If download fails due to network, retry with exponential backoff.'
• Specify constraints for all numeric parameters: max_results (1-100), and for save_path (absolute path required, must have write permissions).
• Translate all descriptions to English or provide English-only versions for LLM compatibility. Keep Chinese as optional secondary documentation.

Inferred effective spec: 2025-06-18+.

• Relies on Stateful initialize / Mcp-Session-Id (removed; protocol is stateless) - make each request self-contained