本文へ移動
cccskills
無料GitHub で公開

life-recorder-setup

Build, install, pair, and operate the Life Recorder iPhone-to-Mac local transcription system. Use when a user asks to reproduce or troubleshoot this project.

browser-use/life-recorder
インストール方法を見る

含まれるファイル（19）

  • SKILL.md2.3 KB
  • .gitignore314 B
  • ios/LifeRecorder.xcodeproj/project.pbxproj15.6 KB
  • ios/LifeRecorder.xcodeproj/project.xcworkspace/contents.xcworkspacedata135 B
  • ios/LifeRecorder/ChunkWriter.swift4.1 KB
  • ios/LifeRecorder/Info.plist1.5 KB
  • ios/LifeRecorder/LifeRecorderApp.swift7.6 KB
  • ios/LifeRecorder/Recorder.swift6.6 KB
  • ios/LifeRecorder/Storage.swift7.7 KB
  • ios/LifeRecorder/UploadManager.swift10.4 KB
  • ios/LifeRecorderTests/NativeUploadTests.swift2.6 KB
  • ios/LifeRecorderTests/RecorderTests.swift3.9 KB
  • ios/project.yml729 B
  • LICENSE1.1 KB
  • README.md3.8 KB
  • receiver/receiver.py15.1 KB
  • receiver/setup.py5.0 KB
  • tests/audio_smoke.py4.5 KB
  • tests/test_receiver.py5.7 KB

SKILL.md（原文）

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Life Recorder setup

Build the native iOS target in ios/LifeRecorder.xcodeproj, configure the Python receiver in receiver/, and keep all runtime state outside the repository.

Phone control

Use Apple CoreDevice (xcrun devicectl) for read-only device inspection, installation, and launching the bundle on a connected device. Use the CUA iPhone Mirroring surface for visual UI actions when the phone is available. Do not bypass a passcode or protected system dialog. If iOS reports that the device is locked, ask the user to unlock it and open Life Recorder once.

Required user permissions

The user must handle or approve:

  • Xcode signing with their Apple developer account and development team
  • Trusting the Mac and enabling Developer Mode on the iPhone
  • Microphone permission for Life Recorder
  • Local-network permission for Life Recorder
  • Any macOS firewall or protected credential prompt

Never request or store the user's Apple password, iPhone passcode, receiver token, private key, or pairing page in the repository.

Setup workflow

  1. Check that Xcode, Python, ffmpeg, whisper.cpp, and a GGML Whisper model are installed.
  2. Build and install the app on the connected iPhone using the user's team and device identifier.
  3. Run receiver/setup.py with a private data directory and --install-agent. Keep the generated pairing page private and delete it after pairing if desired.
  4. Pair the iPhone, switch recording on, and verify a short upload and local Markdown update.
  5. For a stalled queue, inspect the receiver health endpoint and the phone's pending audio count. Opening the app while unlocked reactivates its upload manager; do not force-quit it.
  6. Keep life.md, audio, SQLite state, tokens, certificates, pairing pages, and device-specific validation artifacts outside Git.

The phone records in one-minute chunks and queues them while offline. The Mac receiver transcribes with local Whisper, filters obvious repetitive/stage-direction hallucinations, and writes a continuous Markdown transcript with hourly markers. Reboots and force-quits require one manual app open because iOS does not provide a reliable silent microphone launch at boot.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

ログインしてレビューを書く
このスキルの問題を報告する