Server defines 5 tools with complete input schemas and descriptions. Naming follows verb_noun convention (generate-image, chat, scrape, seo-audit, generate-content). Descriptions are present (avg ~80 chars) but lack actionable context for LLM selection. Parameters have types and constraints (min/max, enums implicit in descriptions). Critical gaps: no output schemas documented, no error handling guidance, no parameter descriptions for nested objects (e.g., 'messages' array items lack field documentation), and no indication of idempotency or side effects. Tool composition is reasonable but 'generate-content' conflates multiple content types into one tool.

• Document output schemas for all tools. Example for generate-image: {"image_url": string, "width": int, "height": int, "seed": int}. Example for chat: {"content": string, "model": string, "tokens_used": int, "stop_reason": string}.
• Expand chat tool's 'messages' parameter schema to explicitly define item structure: {"type": "object", "properties": {"role": {"type": "string", "enum": ["user", "assistant", "system"]}, "content": {"type": "string"}}, "required": ["role", "content"]}.
• Add error recovery hints to all tool descriptions. Example: 'If scrape fails with timeout, try extract_fields=["title", "text"] (fewer fields = faster). If still fails, the URL may be blocked or require authentication.'
• Split generate-content into five separate tools: create_blog_post, create_social_caption, create_script, create_email, create_product_description. Each inherits topic, length, extra_instructions but has a focused name and description.
• Add idempotency and side-effect declarations to descriptions. Example for generate-image: 'Idempotent: same prompt + params = same image (deterministic seed). Safe to retry.' Example for chat: 'Non-idempotent: each call appends to conversation history. Retrying may produce different responses.'

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry