Server has well-structured schemas with Zod validation and comprehensive parameter descriptions in Chinese. All 5 tools have explicit descriptions and input schemas. However, descriptions are domain-specific (image generation API) and lack LLM-optimized guidance on WHEN to use each tool vs alternatives. Parameter descriptions are detailed but lengthy (avg 150+ chars), and output schemas are not documented. Error handling exists but lacks recovery guidance. Tool naming is clear (verb_noun pattern) but composition could be tighter.

• Document output schemas for all tools. E.g., generate_image returns {imageContent: ImageContent, preview: ImagePreview, taskId: string, credits_used: number}. LLMs need to know what fields are available for downstream calls.
• Add LLM-optimized descriptions (50-150 chars) that explain WHEN to use each tool. E.g., estimate_credits: 'Check credit cost before generating. Call this first if user is budget-conscious.' This guides tool selection.
• Shorten parameter descriptions to 10-100 chars and move model-specific limits into enum descriptions or validation errors. E.g., model param description: 'Image generation model (see enum for cost/quality tradeoffs)' instead of 194-char prose.
• Document error categories in tool descriptions. E.g., generate_image: 'Returns error if credits insufficient (user-fixable: add credits), model unsupported (fatal), or timeout (retryable).'
• Add a dry-run mode to generate_image: 'Set dryRun=true to preview cost and parameters without generating. Useful for expensive 4K requests.' Prevents accidental high-cost generations.

Inferred effective spec: 2025-06-18+.

• Current-spec patterns present: structured tool output