A macOS app that captures your screen, processes it with AI in the cloud, and makes your activity searchable through a knowledge base.

macOS App → DO Spaces (raw) → Valkey Queue → DO Function → DO Spaces (processed) → Knowledge Base

The macOS app captures screenshots and uploads them to DigitalOcean. A serverless function processes each image using a vision AI model, extracting text and context. The results are stored in a searchable knowledge base.

See docs/architecture.md for details.

• macOS 14+ (Sonoma or newer)
• Xcode with CLI tools:

  sudo xcode-select -s "/Applications/Xcode.app/Contents/Developer"
  sudo xcodebuild -license accept

• XcodeGen:

  brew install xcodegen

• doctl (for serverless deployment):

  brew install doctl
  doctl auth init

Create these resources in the nyc3 region:

1. Spaces Buckets (2):

   • remember-everything-raw - stores uploaded screenshots
   • remember-everything-processed - stores processed text output

2. Managed Valkey (Basic 1GB tier):

   • Name: remember-everything-queue
   • Acts as the message queue between upload and processing

3. Functions Namespace:

   doctl serverless namespaces create --label remember-everything --region nyc3
   doctl serverless connect remember-everything

4. API Credentials:

   • Spaces access key and secret
   • Valkey connection details (host, port, password)
   • DO API token for Serverless Inferencing

See docs/secrets-reference.md for details.

mkdir -p ~/Library/Application\ Support/RememberEverything
cp macos/Config/config.example.json ~/Library/Application\ Support/RememberEverything/config.json

Edit config.json with your DO credentials:

{
  "captureIntervalSeconds": 10,
  "dataDirectory": "~/RememberEverything/data",
  "spaces": {
    "endpoint": "https://nyc3.digitaloceanspaces.com",
    "region": "nyc3",
    "bucket": "remember-everything-raw",
    "accessKeyId": "YOUR_KEY",
    "secretAccessKey": "YOUR_SECRET"
  },
  "valkey": {
    "host": "your-cluster.valkey.db.ondigitalocean.com",
    "port": 25061,
    "password": "YOUR_PASSWORD",
    "tls": true,
    "listName": "images-to-process"
  },
  "upload": {
    "enabled": true
  }
}

Debug build:

./build_macos.sh

Release build (install to /Applications):

cd macos
xcodegen generate
xcodebuild -project RememberEverything.xcodeproj -scheme RememberEverything -configuration Release -derivedDataPath ../.derived/macos-release
cp -R ../.derived/macos-release/Build/Products/Release/RememberEverything.app /Applications/
open -a /Applications/RememberEverything.app

1. When prompted, open System Settings → Privacy & Security → Screen Recording
2. Enable RememberEverything
3. Restart the app

cp serverless/.env.example serverless/.env
# Edit .env with your credentials

./scripts/deploy_function.sh

1. Go to DO Console → GenAI Platform → Knowledge Bases
2. Create a new KB connected to remember-everything-processed bucket
3. Configure automatic sync
4. Create an Agent using this KB

macOS app logs:

tail -f /tmp/RememberEverything-debug.log

Function logs:

doctl serverless functions get process-images/process --logs

Queue status:

redis-cli -h $VALKEY_HOST -p 25061 --tls -a $VALKEY_PASSWORD LLEN images-to-process

./scripts/install_launch_agent.sh

To uninstall:

./scripts/uninstall_launch_agent.sh

*Cost varies significantly based on model choice and capture interval. Using Llama 3.2 Vision instead of GPT-5 can reduce inference costs by 75%.

• Architecture Overview
• Secrets Reference
• Troubleshooting Guide

remember-everything-ai/
├── macos/                  # macOS capture app (Swift)
│   ├── Capture/           # Screen capture service
│   ├── Config/            # Configuration management
│   ├── Upload/            # Spaces upload + Valkey queue
│   └── ...
├── serverless/            # DO Functions (Python)
│   └── packages/
│       └── process-images/
│           └── process/   # Image processing function
├── scripts/               # Utility scripts
├── docs/                  # Documentation
└── prds/                  # Product requirements