Server has 7 tools with explicit registration, Zod schemas, and tool annotations. Naming follows verb_noun convention (hybrid_*). Descriptions are present (avg 120 chars) but lack actionable context for LLM selection. Parameter schemas are well-typed with enums and constraints, but descriptions are sparse or missing for nested objects. Output schemas are undocumented, LLMs cannot predict response structure. Error handling exists but lacks recovery guidance. No pagination support despite potential for large model lists.

• Document output schemas for all tools. Add a 'returns' field to each tool definition describing the JSON structure, field types, and what each field means. Example: hybrid_status returns {status: 'healthy'|'degraded', activeModel: string, provider: 'ollama'|'client'|'litellm'}.
• Add descriptions to nested parameter objects. The 'options' parameter in hybrid_generate and hybrid_chat should have a description like 'Generation parameters (temperature controls randomness 0-2, top_p controls diversity 0-1, etc.)'.
• Implement pagination for hybrid_list_models. Add limit (default 20, max 100) and offset parameters. Return {models: [...], total: number, hasMore: boolean} so LLMs can iterate through large model lists.
• Enhance error messages with recovery guidance. Wrap tool handlers to catch errors and return structured responses like {error: 'Model not found', suggestion: 'Call hybrid_list_models to see available models', retryable: false}.
• Clarify hybrid_route_anthropic purpose. Expand description to: 'Route Anthropic Messages API requests through the hybrid router, optionally overriding the model or fallback strategy. Use this when you need fine-grained control over routing; use hybrid_chat for simpler chat interactions.'

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

• Relies on Sampling (deprecated) - integrate directly with the LLM provider API