Commands
Full reference for all fma CLI commands.
Global options
fma [--debug] <command>
| Option | Description |
|---|---|
--debug | Enable verbose debug logging. Useful for troubleshooting. |
Example:
fma --debug deploy
fma init
Initialize a new model project.
fma init <model_name>
Creates a directory <model_name>/ with a model.py template and an empty metadata.yaml in the current folder.
Example:
mkdir my-project && cd my-project
fma init image_generator
Result:
my-project/
├── image_generator/
│ └── model.py
└── metadata.yaml
fma login
Authenticate with FlyMyAI.
fma login [--username <username>] [--password <password>]
| Option | Description |
|---|---|
--username | Your FlyMyAI username. Prompted if not provided. |
--password | Your FlyMyAI password. Prompted securely if not provided. |
Credentials are stored locally in ~/.config/fma/config.yaml.
Examples:
# Provide credentials inline
fma login --username alice --password s3cret
# Interactive — prompts for username and password
fma login
fma logout
Log out and clear stored credentials.
fma logout
Removes the auth token and profile from local configuration.
fma deploy
Deploy the current model to FlyMyAI infrastructure.
fma deploy
Must be run from the project root directory (the folder that contains your <model_name>/ folder and metadata.yaml). Requires authentication via fma login first.
On success, the model metadata is written to metadata.yaml.
Example:
cd my-project
fma deploy
# => Successfully deployed model image_generator! ✅
# => Model data was saved in metadata.yaml file
If the model already exists, the CLI will suggest running fma update instead.
fma update
Redeploy the model. This is a shortcut that runs fma delete followed by fma deploy.
fma update
Use this when you've made changes to your model and want to push a new version under the same name.
Example:
# After modifying model.py
fma update
# => Model deleted successfully! ✅
# => Successfully deployed model image_generator! ✅
fma delete
Delete the currently deployed model.
fma delete
Removes the model from FlyMyAI infrastructure. Requires authentication.
Example:
fma delete
# => Model deleted successfully! ✅
fma logs
Fetch startup logs for the deployed model. This is useful for debugging issues during model initialization (e.g. missing dependencies, import errors, crashes in initialize()).
fma logs [--output-file <filename>]
| Option | Description |
|---|---|
--output-file | Save logs to a file instead of printing to the terminal. |
Examples:
# Print logs to terminal
fma logs
# Save logs to a file
fma logs --output-file startup_logs.txt
# => Logs were save into startup_logs.txt.
If the model started successfully, logs may not be available. You will see a message:
Logs are not available. This could happen because project started succesfully.
This is normal — startup logs are primarily available when something goes wrong.
Typical workflow
Here is a complete example of the end-to-end workflow:
# 1. Create and enter a project directory
mkdir my-project && cd my-project
# 2. Initialize the model
fma init text_summarizer
# 3. Edit the model code
# (edit text_summarizer/model.py with your inference logic)
# 4. Log in
fma login
# 5. Deploy
fma deploy
# 6. Check logs if something went wrong
fma logs --output-file logs.txt
# 7. Make changes and redeploy
fma update
# 8. When you're done, clean up
fma delete