Strong schema definitions with proper JSON Schema types, enums, and constraints across all 8 tools. Tool names follow verb_noun convention (generate, edit, list, read, create, append, finalize). Descriptions are present and actionable (avg ~80 chars). All parameters have type definitions and descriptions. Key gaps: descriptions could be more LLM-optimized (some lack 'when to use' context); error handling guidance is minimal; no confirmation pattern for destructive operations (generate/edit/create_upload); output schemas documented in contracts but not visible in server.ts registration.

• Expand tool descriptions to 100-150 chars, adding 'when to use' and 'when NOT to use' guidance. E.g., 'Generate Image: Create new images from text prompts. Use this for original content. For modifications, use Edit Image instead.'
• Add explicit error recovery guidance in descriptions: 'If generation fails, check model availability with List Models, then retry with a different model.'
• Implement dry-run or confirmation pattern for imagegen_generate and imagegen_edit. Return a preview/cost estimate before committing the operation.
• Document output schema in server.ts registration (currently only in contracts). Ensure responses include asset_id, model, provider, and any IDs needed for chaining.
• Add rate-limit and timeout guidance to descriptions. E.g., 'Generation may take 10-60s depending on model and quality_profile. Requests timeout after 5 minutes.'

Inferred effective spec: 2025-06-18+.

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