Single tool with adequate schema but significant gaps in descriptions and parameter documentation. Tool name 'foundation-models' lacks action verb (should be 'generate_text' or similar). Parameter descriptions exist but are minimal (temperature: 18 chars, max_tokens: 34 chars). No output schema documented. Error handling returns plain text rather than structured guidance. Tool description (54 chars) is below the 194-char baseline for production tools. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) present.

• Rename tool from 'foundation-models' to 'generate_text' or 'generate_text_with_foundation_models' to start with an action verb and clarify intent.
• Expand tool description to 150-250 characters: 'Generate text using Apple Foundation Models. Accepts a prompt and optional temperature (0.0-1.0, controls randomness) and max_tokens (output length limit). Returns generated text. Use for creative writing, summarization, Q&A, and code generation.'
• Document output schema in tool description or as a separate field: 'Returns {text: string, tokens_used?: integer, model?: string}' so LLMs know what fields to expect.
• Enhance parameter descriptions: temperature → 'Controls randomness (0.0=deterministic, 1.0=maximum, default 0.7). Lower for factual tasks, higher for creative.'; max_tokens → 'Maximum tokens to generate (1-4096, default varies by model). Limits output length.'
• Implement structured error responses with classification: return {error: string, type: 'retryable'|'user_fixable'|'fatal', suggestion?: string} instead of plain text, enabling LLM recovery logic.

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

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