Datahuman

A native macOS SwiftUI prototype for "Datadog for Humans". This build focuses on UI parity with the provided frames, live input telemetry, BLE heart rate ingestion, and local SQLite storage.

Features implemented

• SwiftUI + AppKit UI with dark neon styling and chart bands
• Hover tooltip, crosshair, and time range/bin selectors
• Input monitoring via event taps (keystroke count, clicks, pointer distance)
• Frontmost app/context via Accessibility
• ScreenCaptureKit stream (no images stored)
• BLE heart rate via CoreBluetooth (standard HR service)
• SQLite persistence (local file in iCloud container if available)
• Wispr Flow dictation stats integration (words dictated, WPM, streaks)

Privacy & data usage (prototype)

This is a prototype that captures sensitive data to power the UI. It counts keystrokes via Input Monitoring event taps (no keycodes or text are captured or stored), collects frontmost app/window context via Accessibility, and streams screen frames via ScreenCaptureKit (frames are processed in-memory; no images are stored). Avoid entering sensitive information while running the app, and contact the maintainers for privacy questions before deploying in any shared or production environment.

Running (SwiftPM)

swift build
swift run

Build macOS app bundle

./scripts/build_app.sh
open dist/Datahuman.app

Note: macOS privacy permissions (Screen Recording, Input Monitoring, Accessibility, etc.) are tied to an app's code signature. If you rebuild an unsigned app, macOS can treat each build as a new app and re-prompt on launch. build_app.sh produces an unsigned app bundle by default. If you have a signing identity, you can codesign:

codesign --force --deep --sign "Apple Development: Your Name (TEAMID)" \
  --entitlements Entitlements/Datahuman.entitlements dist/Datahuman.app

Icon packaging knobs (trade size vs fidelity):

# Include the 1024x1024 rendition in Assets.car (bigger; best quality at very large icon sizes).
APP_ASSETS_CAR_INCLUDE_1024=1 ./scripts/build_app.sh

# Omit Assets.car entirely (smallest build; macOS may fall back to AppIcon.icns).
APP_INCLUDE_ASSETS_CAR=0 ./scripts/build_app.sh

Optional versioning:

./scripts/build_app.sh --version=0.1.0

Build and run immediately:

./scripts/build_app.sh --run

CI / Release

• CI runs on pull requests and pushes to main to build, test, and package the app bundle.
• Release runs on tags matching v* or manual dispatch to build dist/Datahuman-<version>.zip.

Releases (prebuilt app)

GitHub Releases include a zipped .app bundle (Datahuman-<version>.zip).

Note: Release builds are currently not code-signed or notarized, so macOS may block first launch. If that happens, either right-click Datahuman.app and choose Open, or remove the quarantine attribute:

xattr -dr com.apple.quarantine /path/to/Datahuman.app

If you'd rather not run an unsigned prebuilt app, compile from source using SwiftPM (swift run) or ./scripts/build_app.sh.

Testing BLE / WHOOP 5.0 discovery

1. Pair the WHOOP device in macOS Bluetooth settings if needed.
2. Run the app (swift run) and wait for the BLE scan to connect.
3. Inspect the log file: ~/Library/Application Support/Datahuman/ble-debug.jsonl
   • It logs discovered services, characteristics (with properties), and raw value notifications.
   • This file can include sensitive health/biometric data. Treat it as sensitive: restrict file permissions, avoid sending it unencrypted, and delete or redact it when debugging is done. For production builds, disable or limit logging so heart-rate and other biometric values are not retained.
4. If no HR is shown, the app will still log custom notify characteristics so we can identify the right UUIDs.

Permissions

The app will request:

• Input Monitoring
• Accessibility
• Screen Recording
• Bluetooth

For iCloud sync, add the entitlements in Entitlements/Datahuman.entitlements and configure the container identifier in Xcode. The repo ships with a placeholder container (iCloud.com.example.datahuman) so you must replace it with your own iCloud container identifier.

Wispr Flow Integration

If Wispr Flow is installed, Datahuman automatically displays your dictation stats in the "Dictated" stat slot:

• Words dictated - matches the selected time range (1H, TODAY, WEEK, etc.)
• Dictation count - number of dictation sessions in the time range
• WPM avg - your average words per minute
• Day streak - consecutive days of dictation

Stats are read directly from Wispr Flow's SQLite database (~/Library/Application Support/Wispr Flow/flow.sqlite) and update when you change time ranges. Polled every 30 seconds for new dictations. If Wispr Flow is not installed, the stat shows "--" with "Wispr Flow not detected".