sleepy_agent_ios/BUILD.md

Building Sleepy Agent iOS

Prerequisites

1. macOS with Apple Silicon (M1/M2/M3/M4)
2. Xcode 15 or later (from App Store)
3. iOS 15.0+ device or simulator
4. CocoaPods (sudo gem install cocoapods)

Quick Start

Option 1: Xcode GUI (Easiest)

1. Open the Swift Package in Xcode:

   cd sleepy_agent_ios
   open Package.swift
   

2. Xcode will resolve the package and create a project

3. Add LiteRT dependency:

   • File → Add Package Dependencies
   • Search for TensorFlow Lite or LiteRT
   • Or manually download the framework

4. Configure signing:

   • Select the project
   • Signing & Capabilities → Team → Select your Apple ID
   • Bundle Identifier: com.sleepy.agent (or your own)

5. Build and run:

   • Select target device (your iPhone or a simulator)
   • Press Cmd+R

Option 2: CocoaPods + xcodebuild (Command Line)

1. Install CocoaPods dependencies:

   cd sleepy_agent_ios
   pod install
   

2. Open the workspace in Xcode:

   open SleepyAgent.xcworkspace
   

3. Or build from command line:

   make build          # Debug build
   make archive        # Release archive
   

Option 3: Create Fresh Xcode Project

If the above doesn't work, create a new project:

# Make the script executable
chmod +x create-project.sh

# Run the setup script
./create-project.sh

Then manually copy the Swift files from SleepyAgent/ into your new project.

Creating an .ipa File

For Development/Debugging:

1. Archive the app:

   make archive
   

2. Open the archive in Xcode Organizer:

   open build/SleepyAgent.xcarchive
   

3. Export as IPA:

   • In Organizer, select the archive
   • Click "Distribute App"
   • Choose "Development"
   • Follow prompts to export

For Sideloading (No Developer Account):

Use AltStore or similar tools to install the .ipa without a paid developer account.

For Jailbroken Devices:

# Build and package without signing
make build-release
cd build/Release-iphoneos
mkdir -p Payload/SleepyAgent.app
cp -r SleepyAgent.app/* Payload/SleepyAgent.app/
zip -r SleepyAgent.ipa Payload

Troubleshooting

"No such module 'TensorFlowLite'"

LiteRT-LM needs to be added manually. Options:

1. Use CocoaPods: Edit Podfile and add the correct pod
2. Use Swift Package Manager: Add TensorFlow Lite as a dependency
3. Manual framework: Download from TensorFlow releases

Signing Issues

# Check available signing identities
security find-identity -v -p codesigning

# For development builds, use automatic signing in Xcode
# For distribution, you need a paid Apple Developer account ($99/year)

"Could not find iOS SDK"

# Select Xcode command line tools
sudo xcode-select --switch /Applications/Xcode.app/Contents/Developer

# Accept license
sudo xcodebuild -license accept

Model Download Issues

The app downloads models from HuggingFace. If downloads fail:

1. Manually download from: https://huggingface.co/litert-community
2. Place in: Files app → On My iPhone → Sleepy Agent → models/

File Structure

sleepy_agent_ios/
├── SleepyAgent/                    # Source code
│   ├── App/                        # Entry point
│   ├── Core/                       # Models, DI
│   ├── Inference/                  # LLM engine
│   ├── Services/                   # Audio, network, etc.
│   ├── UI/                         # SwiftUI views
│   ├── Info.plist                  # App config
│   └── SleepyAgent.entitlements    # Capabilities
├── Package.swift                   # Swift Package Manager
├── Podfile                         # CocoaPods config
├── Makefile                        # Build automation
└── BUILD.md                        # This file

Build Commands

# Setup
make setup          # Install CocoaPods
make check          # Check environment

# Building
make build          # Debug build (simulator)
make build-release  # Release build (device)
make archive        # Create .xcarchive
make ipa            # Create .ipa

# Installation
make install        # Install to connected device

# Cleanup
make clean          # Clean build artifacts
make clean-all      # Full clean including Pods

Debugging on Device

1. Connect iPhone via USB
2. Trust this computer on iPhone
3. In Xcode: Window → Devices and Simulators
4. Select your device
5. Build and run (Cmd+R)

Notes

• LiteRT-LM iOS models: https://huggingface.co/collections/litert-community/ios-models
• Physical device recommended - Simulator can't test audio/camera well
• First launch - Model loading may take 10-30 seconds depending on size
• RAM requirements - E2B needs ~4GB free, E4B needs ~6GB free

Need Help?

Open an issue at: https://github.com/sleepyeldrazi/sleepy-agent/issues