Server has 3 tools with schemas and descriptions, but quality is inconsistent. oai_generate_image has comprehensive schema with proper enums and constraints (prompt max 32000 chars, n 1-10, output_format enum). However, descriptions are generic and lack LLM-optimized guidance on WHEN to use each tool or what happens on failure. health_check and simple_generate_image lack depth. No error handling guidance, no recovery patterns, no output schema documentation. Parameters are well-typed but descriptions don't explain dependencies or constraints in natural language. Missing: pagination, output structure docs, idempotency guidance, and actionable error messages.

• Expand tool descriptions to 100-200 chars. For oai_generate_image: 'Generate images from text prompts using OpenAI's DALL-E API. Use this for detailed, high-quality images. Returns URLs to generated images. Supports multiple sizes, formats, and quality levels. Requires OPENAI_API_KEY.'
• Document output schema explicitly. E.g., 'Returns: {images: [{url: string, revised_prompt: string}], created: number}' so LLM knows what fields to extract.
• Add error recovery guidance to descriptions. E.g., 'If rate-limited (429), wait 60s and retry. If invalid prompt (400), simplify text. If API key missing, check OPENAI_API_KEY env var.'
• Distinguish oai_generate_image from simple_generate_image in descriptions. E.g., simple_generate_image: 'Quick image generation with minimal options. Use when you only have a text prompt and want defaults.' oai_generate_image: 'Advanced image generation with full control over size, quality, format, and moderation.'
• Add parameter dependency hints. E.g., output_compression: 'Only applies to webp or jpeg formats. Ignored for png. Range 0-100 where 100 is no compression.'