Single tool with minimal definition quality. Tool name 'ocr_image' lacks action verb prefix (should be 'extract_text_from_image' or 'recognize_text_in_image'). Description is 54 chars, acceptable length but lacks WHEN to use it, prerequisites, or output structure guidance. Input schema present but parameter 'file_path' has only basic description; no format constraints, path validation rules, or supported image formats documented. No output schema documented, LLM cannot predict response structure (filename, annotations array with text/confidence/bounding_box fields). Error handling returns dict with 'error' key but provides no recovery guidance or categorization (retryable vs fatal). No tool annotations (readOnlyHint present semantically but not declared). Baseline: 100% of A+ tools document return types; 94% have docstrings; 100% of A+ params have descriptions. This tool meets ~40% of those baselines.

• Rename tool to 'extract_text_from_image' or 'recognize_text_in_image' to start with action verb and clarify intent.
• Document output schema explicitly: 'Returns {filename: string, annotations: [{text: string, confidence: float 0-1, bounding_box: [x, y, width, height]}]}' or similar.
• Add parameter constraints to file_path description: 'Path to image file (PNG, JPEG, TIFF; max 10MB). Relative paths resolved from current directory.'
• Enhance error messages with recovery guidance: 'File not found. Verify path exists and is readable.' For processing errors: 'Image format unsupported or corrupted. Try PNG or JPEG.'
• Expand tool description to 100-150 chars: 'Extract text from images using macOS Vision OCR. Returns recognized text with confidence scores and bounding boxes. Requires macOS 10.15+. Supports PNG, JPEG, TIFF.'