CLI Command
moro init
Scaffolds a new project directory structure, creates the local SQLite state database, and writes an initial moro.yaml specification.
Usage
moro init [PROJECT_NAME] [OPTIONS]
Arguments & Options
| Parameter | Type | Default | Description |
|---|---|---|---|
| PROJECT_NAME | Argument (String) | moro-project | Name of the directory to create. |
| --template, -t | Option (String) | default | Starter template: default, support-bot, json-extractor, or code. |
| --model, -m | Option (String) | Qwen/Qwen2.5-1.5B | Default foundation model to pre-populate in moro.yaml. |
Created Directory Layout
my-project/ ├── moro.yaml # Project configuration & hyperparameter recipe ├── .moro/ # Local metadata, DAG state manager, and SQLite DB │ └── moro.db ├── data/ │ ├── raw/ # Place your raw .jsonl, .csv, or .parquet inputs here │ └── compiled/ # Output of 'moro data build' with MI Guard tags ├── runs/ # Training run artifacts, checkpoints, and loss curves ├── releases/ # Exported GGUF models, Modelfiles, and signed SBOMs └── eval/ # Evaluation test suites and passing criteria
Example
# Initialize a customer support project with Llama 3.2 moro init support-foundry --template support-bot --model meta-llama/Llama-3.2-1B cd support-foundry