ORO CLI

Command line tool for OpenRobOps (ORO): manage configuration as code and query your fleet from the terminal, against one or more ORO instances.

oro use myinstance
oro apply -f missions/
oro get config --kind ActionDefinition
oro get robots

Install

Requires Python 3.12 or newer.

pip install oro-cli

Contexts

A context is one ORO instance plus the API key used to talk to it. Create an API key in your ORO instance (user menu > API keys), then register the instance:

oro context add myinstance --url https://oro.example.com/api --api-key oro_xxx

The first context you add becomes the current one. Add more and switch between them:

oro context add local --url http://localhost:3000/api --api-key oro_yyy --verbose
oro use local                 # same as: oro context use local
oro context list              # all contexts, current marked with *
oro context current           # name of the current context
oro context remove local
oro whoami                    # current context, URL, and whether the key authenticates

Contexts are stored in ~/.config/oro/config.yaml (or $XDG_CONFIG_HOME/oro/config.yaml), readable only by you. API keys are stored in plain text. The file looks like this:

current: myinstance
contexts:
  myinstance:
    url: https://oro.example.com/api
    api_key: oro_xxx
    verbose: false
  local:
    url: http://localhost:3000/api
    api_key: oro_yyy
    verbose: true

You can edit it by hand, but prefer the oro context commands: a file that does not parse makes every command fail until it is fixed. verbose: true logs every HTTP request and response for that context, the same as passing --verbose.

Environment variables

For CI and scripts, environment variables override the file:

When both ORO_URL and ORO_API_KEY are set, no config file is needed. Setting only one of them overrides that field of the current context. Empty values count as unset.

Commands

Configuration as code

oro apply -f config.yaml           # show a diff per object, ask, then apply
oro apply -f config.yaml -y        # apply without asking
oro apply -f config.yaml --compare # only show differences
oro apply -f configs/              # every .yaml, .yml and .json file under the directory
oro apply -f configs/ --compare --all              # also list server objects missing locally
oro apply -f configs/ --compare --all --exclude Test

oro get config --kind ActionDefinition             # full objects as YAML
oro get config --kind ActionDefinition wave        # one object
oro get config --kind ActionDefinition --json      # as JSON
oro get config --kind ActionDefinition --summary   # short table
oro get config --kind ActionDefinition --all       # include normally hidden elements
oro list kinds                                     # available kinds

oro delete config --kind ActionDefinition wave
oro delete -f config.yaml                          # delete every object in the file

Config files hold one or more objects as YAML documents (or a JSON list):

kind: ActionDefinition
metadata:
  id: wave
spec:
  label: Wave
  type: PublishToTopic

apiVersion is filled in when missing. A metadata.scope field is ignored with a warning, so files written for other tools keep working.

Robots

oro get robots        # name, id, agent version
oro describe robots   # also online state and last seen

Global options

oro --verbose <command>   # log HTTP traffic to stderr
oro --version
oro --help

Development

To try the tool from a checkout without publishing it, install it in editable mode. The oro command then runs the code in your working tree, so edits take effect immediately:

git clone https://github.com/openrobops/oro-cli.git
cd oro-cli
python3 -m venv .venv && source .venv/bin/activate
pip install -e '.[dev]'

oro --version
oro context add local --url http://localhost:3000/api --api-key oro_xxx
oro whoami

Or, without a virtualenv, run it straight from the repo with uv or pipx: uv run oro --help or pipx install -e ..

Tests and lint:

pytest
ruff check . && ruff format --check .

Release: bump version in pyproject.toml, commit, then publish a GitHub release with a vX.Y.Z tag. The publish workflow builds the package and uploads it to PyPI through trusted publishing (no token needed). To publish by hand instead:

python -m build
twine upload dist/*

License

Apache 2.0. See LICENSE.