trossen_cloud_cli

CLI for interacting with Trossen Cloud datasets, models, and training jobs.

Installation

Requires Python 3.11+.

uv tool install trossen-cloud-cli

Or with pipx:

pipx install trossen-cloud-cli

Or with pip:

pip install trossen-cloud-cli

Authentication

Create an API token in the Trossen Cloud web UI, then:

# Directly provide your token
trc auth login --token <your-api-token>

# Or receive a password prompt
trc auth login

# Check authentication status
trc auth status

The token is stored securely in your OS keyring.

Usage

Datasets

# Upload a local dataset
trc dataset upload ./my-data --name my-dataset

# List episodes in a dataset
trc dataset episodes <dataset-id>

# Add a single new episode to an existing dataset
trc dataset add-episodes <dataset-id> ./episode_000042.mcap

# Add multiple episodes at once by pointing at a directory
# (every .mcap inside is uploaded; non-.mcap files are skipped with a warning)
trc dataset add-episodes <dataset-id> ./new-episodes/

# If another edit is already in progress on the dataset, cancel it and retry
# (this discards that edit, which may belong to another process or session)
trc dataset add-episodes <dataset-id> ./new-episodes/ --cancel-in-progress

# Remove episodes from a dataset (by filename)
trc dataset remove-episodes <dataset-id> episode_000042.mcap

# Remove multiple episodes at once (space-separated, up to 200 per call)
trc dataset remove-episodes <dataset-id> episode_000042.mcap episode_000043.mcap episode_000044.mcap

# The .mcap extension is optional, and --force skips the confirmation prompt
trc dataset remove-episodes <dataset-id> episode_000042 episode_000043 --force

# Download a dataset
trc dataset download <dataset-id> ./output

# Browse and manage
trc dataset list --mine
trc dataset info <dataset-id>
trc dataset view <user>/<name>
trc dataset update <dataset-id> --name new-name --privacy public
trc dataset delete <dataset-id>

Models

# Upload a model
trc model upload ./my-model --name my-model

# Download a model
trc model download <model-id> ./output

# Browse and manage
trc model list --mine
trc model info <model-id>
trc model view <user>/<name>
trc model update <model-id> --name new-name
trc model delete <model-id>

Training Jobs

# Create a training job
trc training-job create --name my-job --base-model-id <id> --dataset-id <id>

# Monitor and manage
trc training-job list
trc training-job info <job-id>
trc training-job cancel <job-id>
trc training-job models <job-id>

Run trc usage to see a quick-reference of all commands, or trc <command> --help for detailed help on any command.

Configuration

Transfer settings (chunk sizes, concurrency) can be tuned via trc config. See docs/configuration.md for details.