SKILL.md
iOS Computer Vision Router
You MUST use this skill for ANY computer vision work using the Vision framework.
When to Use
Use this router when:
- Analyzing images or video
- Detecting objects, faces, or people
- Tracking hand or body pose
- Segmenting people or subjects
- Lifting subjects from backgrounds
- Recognizing text in images (OCR)
- Detecting barcodes or QR codes
- Scanning documents
- Using VisionKit or DataScannerViewController
Routing Logic
Vision Work
Implementation patterns → $axiom-vision
- Subject segmentation (VisionKit)
- Hand pose detection (21 landmarks)
- Body pose detection (2D/3D)
- Person segmentation
- Face detection
- Isolating objects while excluding hands
- Text recognition (VNRecognizeTextRequest)
- Barcode/QR detection (VNDetectBarcodesRequest)
- Document scanning (VNDocumentCameraViewController)
- Live scanning (DataScannerViewController)
- Structured document extraction (RecognizeDocumentsRequest, iOS 26+)
API reference → $axiom-vision-ref
- Complete Vision framework API
- VNDetectHumanHandPoseRequest
- VNDetectHumanBodyPoseRequest
- VNGenerateForegroundInstanceMaskRequest
- VNRecognizeTextRequest (fast/accurate modes)
- VNDetectBarcodesRequest (symbologies)
- DataScannerViewController delegates
- RecognizeDocumentsRequest (iOS 26+)
- Coordinate conversion patterns
Diagnostics → $axiom-vision-diag
- Subject not detected
- Hand pose missing landmarks
- Low confidence observations
- Performance issues
- Coordinate conversion bugs
- Text not recognized or wrong characters
- Barcodes not detected
- DataScanner showing blank or no items
- Document edges not detected
Decision Tree
- Implementing (pose, segmentation, OCR, barcodes, documents, live scanning)? → vision
- Need API reference / code examples? → vision-ref
- Debugging issues (detection failures, confidence, coordinates)? → vision-diag
Anti-Rationalization
| Thought | Reality |
|---|---|
| "Vision framework is just a request/handler pattern" | Vision has coordinate conversion, confidence thresholds, and performance gotchas. vision covers them. |
| "I'll handle text recognition without the skill" | VNRecognizeTextRequest has fast/accurate modes and language-specific settings. vision has the patterns. |
| "Subject segmentation is straightforward" | Instance masks have HDR compositing and hand-exclusion patterns. vision covers complex scenarios. |
Critical Patterns
vision:
- Subject segmentation with VisionKit
- Hand pose detection (21 landmarks)
- Body pose detection (2D/3D, up to 4 people)
- Isolating objects while excluding hands
- CoreImage HDR compositing
- Text recognition (fast vs accurate modes)
- Barcode detection (symbology selection)
- Document scanning with perspective correction
- Live scanning with DataScannerViewController
- Structured document extraction (iOS 26+)
vision-diag:
- Subject detection failures
- Landmark tracking issues
- Performance optimization
- Observation confidence thresholds
- Text recognition failures (language, contrast)
- Barcode detection issues (symbology, distance)
- DataScanner troubleshooting
- Document edge detection problems
Example Invocations
User: "How do I detect hand pose in an image?" → Invoke: $axiom-vision
User: "Isolate a subject but exclude the user's hands" → Invoke: $axiom-vision
User: "How do I read text from an image?" → Invoke: $axiom-vision
User: "Scan QR codes with the camera" → Invoke: $axiom-vision
User: "How do I implement document scanning?" → Invoke: $axiom-vision
User: "Use DataScannerViewController for live text" → Invoke: $axiom-vision
User: "Subject detection isn't working" → Invoke: $axiom-vision-diag
User: "Text recognition returns wrong characters" → Invoke: $axiom-vision-diag
User: "Barcode not being detected" → Invoke: $axiom-vision-diag
User: "Show me VNDetectHumanBodyPoseRequest examples" → Invoke: $axiom-vision-ref
User: "What symbologies does VNDetectBarcodesRequest support?" → Invoke: $axiom-vision-ref
User: "RecognizeDocumentsRequest API reference" → Invoke: $axiom-vision-ref