Server has 4 tools with minimal schema and description quality. Three tools have empty input schemas ({}), making them impossible for LLMs to understand parameter requirements. The one tool with a parameter (mobu_scene__save_scene) has a schema but lacks descriptions for what 'path' means in context. Tool descriptions are present but generic (10-50 chars), below the 50-200 char baseline for LLM-optimized descriptions. No output schemas documented. No error handling guidance. No tool annotations (readOnlyHint/destructiveHint). Naming is acceptable (verb_noun pattern) but descriptions fail to explain WHEN to use each tool or what it returns.

• Expand all tool descriptions to 50-200 characters. Example: 'mobu_diagnostics__ping' → 'Verify MotionBuilder is running and return the host system info and embedded Python version. Use this first to confirm connectivity before calling other tools.'
• Document output schemas for all tools. Example: mobu_scene__inspect_scene should return {scene_name: string, model_count: int, frame_rate: float, ...}. This enables LLMs to chain calls.
• Add parameter descriptions to mobu_scene__save_scene. 'path' needs: 'Absolute file path where the scene will be saved. Must end with .fbx extension. Example: /home/user/scenes/my_scene.fbx. Will overwrite existing files.'
• Add tool annotations: mark mobu_scene__save_scene with destructiveHint=true (overwrites files) and idempotentHint=false (repeated calls may produce different results if scene changes). Mark read-only tools with readOnlyHint=true.
• Add error handling to save_scene description: 'If the path is invalid or inaccessible, the tool will return an error with the reason. Ensure the directory exists and you have write permissions.'