vision-framework — Claude Code skill: computer vision for iOS
A skill for adding text recognition, face detection and barcode scanning to iOS apps. Useful when you need OCR or live camera scanning.
npx skills add dpearson2699/swift-ios-skills --skill vision-framework -a claude-codeWhat vision-framework does
The skill covers Vision on iOS: text recognition, face and landmark detection, barcodes and QR codes, image segmentation, object tracking, document scanning. There is a Core ML section too, for running your own models through CoreMLRequest and VNCoreMLRequest.
Two API generations are documented. Modern Swift types such as RecognizeTextRequest and DetectFaceRectanglesRequest use try await request.perform(on:) and start at iOS 18. The legacy layer on VNRequest and VNImageRequestHandler stays relevant when your deployment target is lower. The skill suggests keeping legacy code in separate files or sections.
Vision coordinates are normalized with the origin at the bottom-left, so convert them to UIKit before drawing. For live camera scanning there is DataScannerViewController from VisionKit. Structured document reading with tables and lists arrived in iOS 26.
What it helps with
- pull text out of a photo with confidence scores
- find faces and landmarks in an image
- read a QR or barcode from a camera frame
- cut a person out of the background
- run a custom Core ML model through Vision
Example prompts for the agent
- add Vision OCR to my scanning screen
- I need DataScannerViewController for live QR reading
- convert a face bounding box to UIKit coordinates
How to install vision-framework
1Skills CLI (Vercel)
npx skills add dpearson2699/swift-ios-skills --skill vision-framework -a claude-code2Manually: put the files in place
~/.claude/skills/vision-framework/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/vision-framework/
1Skills CLI (Vercel)
npx skills add dpearson2699/swift-ios-skills --skill vision-framework -a codex2Manually: put the files in place
.agents/skills/vision-framework/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add dpearson2699/swift-ios-skills --skill vision-framework -a cursor2Manually: put the files in place
.cursor/skills/vision-framework/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install dpearson2699/swift-ios-skills vision-framework2Manually: put the files in place
.github/skills/vision-framework/SKILL.md1Skills CLI (Vercel)
npx skills add dpearson2699/swift-ios-skills --skill vision-framework -a gemini-cli2Platform CLI
gemini skills install https://github.com/dpearson2699/swift-ios-skills --path skills/vision-frameworkManually: put the files in place
.devin/skills/vision-framework/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add dpearson2699/swift-ios-skills --skill vision-framework -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about vision-framework
How do I install vision-framework in Claude?
Run npx skills add dpearson2699/swift-ios-skills --skill vision-framework -a claude-code. The skill file lands at ~/.claude/skills/vision-framework/SKILL.md.
What does vision-framework do?
It documents computer vision work on iOS: OCR, face detection, barcodes, segmentation, object tracking and document scanning. Custom Core ML inference gets its own section.
Which iOS version does the modern API need?
Swift-native request types and perform(on:) require iOS 18. For older targets the skill points to the legacy VNRequest pattern.
