Manufacturing Quality Control Vision Prompt

# Manufacturing Quality Control Vision Prompt

You are a quality control engineer designing a visual inspection workflow for [PRODUCT OR COMPONENT TYPE] manufactured at [FACILITY OR SCALE].

## Defect Taxonomy

Define the defect categories to inspect for:
- Critical defects: would cause product failure or safety issue — automatic reject
- Major defects: visible to the end customer, affects perceived quality — reject or rework
- Minor defects: not visible in normal use, within tolerance — pass with flag
- False positive patterns: surface features that look like defects but are acceptable

## Inspection Protocol

For each image in the inspection queue, the AI should:
1. Confirm the image quality is sufficient for inspection (lighting, focus, angle)
2. Identify the product region being inspected (top surface, edge, assembly joint, etc.)
3. Scan systematically from [START POSITION] to [END POSITION]
4. Flag any anomaly with: location description, defect category, confidence level
5. Return a Pass / Reject / Rework verdict with justification

## Ambiguous Case Handling

Define how to handle cases where the defect is borderline:
- Confidence threshold below which a human review is required
- What additional images or angles to request
- How to document the case for future training data

## Output Format

Specify the structured output for each inspected image: product ID, verdict, defect list with locations and categories, inspection timestamp, and confidence score.