the-curator

Turning memories into mastery

Get started

Prerequisites

• Python 3.10 (client) and 3.12 (server)
• Docker
• Node v22

Local setup (development)

Due to different dependencies (and their relations), client (only CLI application) and server need different Python versions. But both use a shared pre-commit setup to enable style consistency in the repository. The configuration is stored in .pre-commit-config.yaml. Install the recommended extensions from .vscode/extensions.json.

Server

Use Python 3.12

python -m venv .venv.server
# Activate envionment based on system (Mac: source .venv.server/bin/activate)
pip install -r requirements-server.txt
pip install -r requirements-local.txt
cp .env.db.example .env.db
cp .env.example .env
pre-commit install

Finally, set the defined variables in .env and .env.db accordingly. For an example of the .env, see here.

To run LlamaIndex components, start the respective services from the compose.yml file:

docker compose up -d vector-db pgadmin

Client (CLI application)

Use Python 3.10

python -m venv .venv
# Activate envionment based on system (Mac: source .venv/bin/activate)
pip install -r requirements-client.txt
pip install -r requirements-local.txt
cp .env.local.example .env.local
pre-commit install

Finally, set the defined variable in .env.local accordingly. If you don't know the value of the AUDIO_DEVICE, leave it empty, the CLI will print a list of available devices on startup.

Client (Web application)

Use Node v22

cd src/frontend
npm install
cp .env.local.example .env.local

Set the defined variable in src/frontend/.env.local accordingly. For an example, see here.

Start

Note

Due to loading the LlamaIndex setup, the startup can take some time. See the logs and wait until the setup is completed.

Terminal

Important

Requires setup for development described here

Use startup files directly:

1. run.py starts Flask server
2. run-cli.py starts CLI application (use defined flags for startup)

For the web client, use npm run dev in the src/frontend directory.

VSCode run configurations

Important

Requires setup for development described here

Run server and client application via the Run and Debug tab

Docker

You can alternatively build and run the server and the web client also in a Docker container. Use compose.yml for the setup. For an example of the .env and src/frontend/.env.local file, see here.

docker compose up -d --build

Access

Development setup: http://127.0.0.1:3050

Docker: http://127.0.0.1:7070

Additional notes

When using the LMStudioLLM, you need to setup 'Meta-Llama-3.1-8B-Instruct' in your local LM Studio installation.

Examples for .env files

Development setup

.env

LECTURE_TRANSLATOR_TOKEN="<TOKEN>"
LECTURE_TRANSLATOR_URL="<URL>"
LLM_URL="<URL>"
ANKI_COLLECTION_PATH="./data/anki_env"
HUGGING_FACE_TOKEN="<TOKEN>"
FRONTEND_URL="http://127.0.0.1:3050"
ANKI_RELATIVE_DATA_DIR="./data/anki_files"
LLM_TO_USE="hosted"  # use local when LM Studio is running, "hosted" will use the defined "LLM_URL" alongside a HuggingFace InferenceClient

src/frontend/.env.local

NEXT_PUBLIC_BACKEND_URL="http://127.0.0.1:5000"

Docker setup

.env

LECTURE_TRANSLATOR_TOKEN="<TOKEN>"
LECTURE_TRANSLATOR_URL="<URL>"
LLM_URL="<URL>"
ANKI_COLLECTION_PATH="/flask-app/data/anki_env"
HUGGING_FACE_TOKEN="<TOKEN>"
FRONTEND_URL="http://127.0.0.1:7070"
ANKI_RELATIVE_DATA_DIR="/flask-app/data/anki_files"
LLM_TO_USE="hosted"

src/frontend/.env

NEXT_PUBLIC_BACKEND_URL="http://127.0.0.1:7071"