# DSH Remote App Review Notes

## Product summary

DSH Remote is a native SwiftUI companion that controls an AI coding Harness running on a computer owned or managed by the user. The iOS app does not download or execute code, expose a terminal, install plugins, edit model credentials, or purchase digital services. Work executes on the user's computer.

The app is an independent open-source project and is not affiliated with DeepSeek AI or Tailscale. The name, icon, About screen, and store copy must preserve that distinction.

## Review without external hardware or accounts

No login or review credentials are required.

1. Launch the app.
2. On the no-computer screen, tap **先体验一下** below the connect button.
3. Open **登录流程上线检查** (`Login flow release check`).
4. Inspect the native conversation and tool-summary cards.
5. Answer the visible confirmation question.
6. Send a prompt; the demo simulates running and completion without network access or an LLM.
7. The `@` picker and subagent catalog/history use offline demo data. The image draft can be exercised with any image explicitly chosen in the system picker or pasted from the clipboard; the demo still makes no network or model call and requests no broad photo-library access.

The offline experience is presented as an explicit secondary action when no computer has been saved and contains no hidden gestures.

## Live connection

Live mode connects directly to the user's Harness computer through Tailscale Serve, an authenticated same-Wi-Fi endpoint, or another HTTPS endpoint the user controls. The same-Wi-Fi endpoint is the recommended first pairing path, is opt-in, is restricted to the Remote API allowlist, and is protected by a high-entropy random bearer credential carried in the pairing QR code. It uses local HTTP and is labeled for trusted private Wi-Fi only. Tailscale is presented only as an optional cross-network or untrusted-network path and is not required by the app. The maintainer operates no relay. No Tailscale account is created or purchased in the app.

## Data and privacy

- The maintainer receives no prompts, code, credentials, diagnostics, analytics, or account data from the iOS app.
- Saved computer names, URLs, and LAN pairing credentials use `UserDefaults` only for app functionality; the privacy manifest declares reason `CA92.1`.
- Camera access occurs only after the user chooses QR scanning.
- Image access occurs only after the user chooses the system photo picker or explicitly pastes an image. Selected images are processed in memory and sent directly to the user's computer; the app has no photo library, relay, or maintainer-operated upload service.
- Local notifications report task completion or an interaction waiting while monitoring is active.
- Prompts can be sent by the user's computer to the model provider configured on that computer. App Store privacy answers and review notes must disclose this user-directed flow without claiming that no third party ever processes content.

## Store submission checklist

- Host the privacy policy and support page at stable public HTTPS URLs before submission.
- Use those URLs in App Store Connect and verify they open without authentication.
- Complete the App Privacy questionnaire consistently with the shipped binary and this policy.
- Use the Developer Tools category and explain the companion relationship in Review Notes.
- Attach a short review video only if Apple requests help reproducing a live-computer connection; the built-in demo should remain sufficient for core review.
- Verify the final archive contains `PrivacyInfo.xcprivacy`, the independent 1024×1024 icon, camera/local-network purpose strings, and no `WKWebView` remote UI.
- Confirm all repository licenses and attribution notices are current before the first public binary.

## Distribution before App Store release

The first external build has passed Beta App Review and the public TestFlight
group is available at <https://testflight.apple.com/join/7Ew6Yk9V>. Developers
may also run the source on their own device through Xcode signing.

TestFlight availability must remain distinct from App Store availability. The
app has not yet been submitted for public App Store distribution.