cli

The following documentation is for the agentos command line utility, which is built using Click. The documentation below is generated automatically by sphinx-click.

agentos

agentos [OPTIONS] COMMAND [ARGS]...

Options

--version: Show the version and exit.

clear-cache

This command clears all virtual environments AND git repos that have been cached by AgentOS on your local file system.

agentos clear-cache [OPTIONS]

Options

-y, --assume-yes: Automatically answers ‘yes’ to all user prompts.

clear-env-cache

This command clears all virtual environments that have been cached by AgentOS in your local file system. All the virtual environments can be automatically recreated when re-running a Component that has requirements_path specified.

agentos clear-env-cache [OPTIONS]

Options

-y, --assume-yes: Automatically answers ‘yes’ to all user prompts.

clear-repo-cache

This command clears all git repos that have been cached by AgentOS on your local file system. These repos will be recreated as you run Components that require them.

agentos clear-repo-cache [OPTIONS]

Options

-y, --assume-yes: Automatically answers ‘yes’ to all user prompts.

freeze

Creates a version of registry_file for Component component_name where all Components in the dependency tree are associated with a specific git commit. The resulting registry_file can be run on any machine with AgentOS installed.

The requirements for pinning a Component spec are as follows:

All Components in the dependency tree must be in git repos
Those git repos must have GitHub as their origin
The current local branch and its counterpart on origin are at the same commit
There are no uncommitted changes in the local repo

agentos freeze [OPTIONS] COMPONENT_NAME

Options

-r, --registry-file <registry_file>: Path to registry file (components.yaml).

-f, --force: Force freeze even if repo is in a bad state.

--use-auto-env, --use-outer-env: If –use-auto-venv is passed, then AgentOS will automatically create a virtual environment under which the Component DAG will be run. If –use-outer-env is passed, AgentOS will not create a new virtual environment for the Component DAG, instead running it in the existing outer Python environment.

Arguments

COMPONENT_NAME: Required argument

init

Initialize current (or specified) directory as an AgentOS agent.

Arguments:

[OPTIONAL] DIR_NAMES zero or more space separated directories to

initialize. They will be created if they do

not exist.

Creates an agent main.py file, a conda env, and an MLflow project file in all directories specified, or if none are specified, then create the files in current directory.

agentos init [OPTIONS] DIR_NAMES

Options

-n, --agent-name <AGENT_NAME>: This is used as the name of the MLflow Project and Conda env for all Directory Agents being created. AGENT_NAME may not contain ‘ ‘, ‘:’, or ‘/’.

Arguments

DIR_NAMES: Optional argument(s)

publish

This command pushes the spec for component component_name (and all its sub-Components) to the AgentOS server. This command will fail if any Component in the dependency tree cannot be frozen.

agentos publish [OPTIONS] COMPONENT_NAME

Options

-r, --registry-file <registry_file>: Path to registry file (components.yaml).

-f, --force: Force freeze even if repo is in a bad state.

--use-auto-env, --use-outer-env: If –use-auto-venv is passed, then AgentOS will automatically create a virtual environment under which the Component DAG will be run. If –use-outer-env is passed, AgentOS will not create a new virtual environment for the Component DAG, instead running it in the existing outer Python environment.

Arguments

COMPONENT_NAME: Required argument

publish-run

agentos publish-run [OPTIONS] ENTITY_ID

Arguments

ENTITY_ID: Optional argument

rerun

agentos rerun [OPTIONS] RUN_ID

Arguments

RUN_ID: Required argument

run

agentos run [OPTIONS] COMPONENT_NAME

Options

-r, --registry-file <registry_file>: Path to registry file (components.yaml).

--entry-point <ENTRY_POINT>: A function of the component that AgentOS Runtime will call with the specified argument set.

-A, --arg-set-list <NAME=VALUE>: A argument for the run, of the form -A name=value. All arguments will be passed to the entry_point function using a Python kwargs-style keyword argument https://docs.python.org/3/glossary.html#term-argument

--arg-set-file <PARAM_FILE>: A YAML file containing arguments for the entry point being run. Will be passed to the entry_point function, along with individually specified args, via a Python kwargs-style keyword argument https://docs.python.org/3/glossary.html#term-argument

--use-auto-env, --use-outer-env: If –use-auto-venv is passed, then AgentOS will automatically create a virtual environment under which the Component DAG will be run. If –use-outer-env is passed, AgentOS will not create a new virtual environment for the Component DAG, instead running it in the existing outer Python environment.

Arguments

COMPONENT_NAME: Required argument

status

ENTITY_ID can be a Component name or a Run ID.

agentos status [OPTIONS] ENTITY_ID

Options

-r, --registry-file <registry_file>: Path to registry file (components.yaml).

--use-auto-env, --use-outer-env: If –use-auto-venv is passed, then AgentOS will automatically create a virtual environment under which the Component DAG will be run. If –use-outer-env is passed, AgentOS will not create a new virtual environment for the Component DAG, instead running it in the existing outer Python environment.

Arguments

ENTITY_ID: Optional argument