Build a Pose Estimation App
with Claude & QuickPose
AI coding assistants like Claude can take you from idea to a working AI fitness app in hours — if the SDK underneath is well-documented and predictable. QuickPose is built exactly that way: a typed Swift API, open sample repos, and public docs that Claude can read and reason about.
An SDK designed for humans reads well to AI too
AI assistants write their best code against APIs with clear types, public documentation and working examples. QuickPose gives Claude all three.
Public, Crawlable Docs
Our documentation at docs.quickpose.ai is fully public — paste a link into your Claude conversation and it can read the exact API surface it's coding against.
Typed Swift API
Features like .fitness(.squats) and QuickPoseThresholdCounter are strongly typed — the compiler catches AI hallucinations before your users do.
Working Sample Apps
Open-source demos on GitHub — push-up counter, squat counter, range of motion — give Claude real reference implementations to pattern-match against.
24 Pre-Built Exercises
Claude doesn't need to invent rep-counting logic — it wires up counters we've already built and tested in production.
One SDK, No Backend
100% on-device processing means no server, no API keys to secure, no infrastructure for your AI assistant to misconfigure — the whole app fits in one Xcode project.
Fast Feedback Loop
Build, run on your iPhone, see the skeleton overlay live. Tight iteration loops are where AI-assisted development shines.
From prompt to working app
Get Your Free SDK Key
Sign up at dev.quickpose.ai — free for up to 10 monthly active devices, no card required.
Open Claude
Use Claude Code in your terminal for full-project work, or claude.ai for planning and snippets.
Paste the Starter Prompt
Our prompt below points Claude at the right docs and repos, and tells it exactly what to build.
Iterate & Ship
Run on your iPhone, refine with follow-up prompts, and submit to the App Store when it feels right.
The starter prompt
Paste this into Claude Code from an empty folder (or adapt it for your existing app). Pick your platform, swap in your own exercise and branding.
Build me a native iOS app in SwiftUI that counts squat reps in real time using the QuickPose iOS SDK. Context you should read first: - SDK repo & README: https://github.com/quickpose/quickpose-ios-sdk - Docs: https://docs.quickpose.ai/docs/MobileSDK - Exercise library: https://docs.quickpose.ai/docs/MobileSDK/ExerciseLibrary Requirements: 1. Integrate the QuickPose iOS SDK via Swift Package Manager. 2. Use the front camera with a full-body skeleton overlay. 3. Use the .fitness(.squats) feature with QuickPoseThresholdCounter to count reps, and show the live count in large text. 4. Add a start/stop button and a session summary screen (total reps, session duration). 5. Read my SDK key from a Config.plist entry named QUICKPOSE_SDK_KEY (I'll add the value myself — never hardcode it). 6. Include the camera-permission text in Info.plist. My SDK key is registered for bundle ID: com.example.squatcounter (change the project to use that bundle ID). Follow the official sample apps' integration pattern rather than inventing your own camera pipeline. When you're done, tell me exactly how to run it on my iPhone from Xcode.
Build me a native Android app in Kotlin (Jetpack Compose) that counts squat reps in real time using the QuickPose Android SDK. Context you should read first: - SDK repo & README: https://github.com/quickpose/quickpose-android-sdk - Docs: https://docs.quickpose.ai/docs/MobileSDK - Exercise library: https://docs.quickpose.ai/docs/MobileSDK/ExerciseLibrary Requirements: 1. Integrate the QuickPose Android SDK via Gradle, following the README's installation steps exactly. 2. Use the front camera with a full-body skeleton overlay. 3. Use the squats fitness feature with the SDK's threshold counter to count reps, and show the live count in large text. 4. Add a start/stop button and a session summary screen (total reps, session duration). 5. Read my SDK key from local.properties via BuildConfig as QUICKPOSE_SDK_KEY (I'll add the value myself — never hardcode it). 6. Add the CAMERA permission and runtime permission request. My SDK key is registered for application ID: com.example.squatcounter (set the project to use that application ID). Follow the official sample apps' integration pattern rather than inventing your own camera pipeline. When you're done, tell me exactly how to run it on my phone from Android Studio.
Build me a React Native app (TypeScript) that counts squat reps in real time using the QuickPose SDK via a native module bridge. Context you should read first: - Official React Native example (follow its bridging pattern): https://github.com/quickpose/quickpose-react-native-pose-estimation - iOS SDK: https://github.com/quickpose/quickpose-ios-sdk - Android SDK: https://github.com/quickpose/quickpose-android-sdk - Docs: https://docs.quickpose.ai/docs/MobileSDK Requirements: 1. Start from the example repo's native module pattern — QuickPose runs natively on each platform; React Native renders the UI. 2. Use the front camera with a full-body skeleton overlay. 3. Use the squats fitness feature with the SDK's threshold counter, and stream the live rep count to the React Native UI. 4. Add a start/stop button and a session summary screen (total reps, session duration). 5. Read my SDK key from react-native-config as QUICKPOSE_SDK_KEY (I'll add the value myself — never hardcode it). 6. Handle camera permissions on both platforms. My SDK key is registered for bundle/application ID: com.example.squatcounter (use it on both platforms). Get iOS working first, then Android. When you're done, tell me exactly how to run it on a real device with npx react-native run-ios and run-android.
.fitness(.jumpingJacks) — see all 24 in the Exercise Library.Things people build with Claude + QuickPose
Office break workout app
Three-exercise micro-workouts with rep counting and a daily streak.
Morning yoga checker
Pose detection with gentle on-screen guidance for a five-pose flow.
Rehab exercise companion
Range-of-motion tracking with weekly progress charts for recovery.
Boxing combo trainer
Real-time movement detection with audio cues to keep the tempo up.
AI-assisted building, answered
Your first rep counter is one prompt away
Grab a free SDK key, paste the starter prompt into Claude, and see your skeleton overlay running today.