ClearML Agent Environment Variables

This page lists the available environment variables for configuring ClearML Agent.

In addition to the environment variables listed below, ClearML also supports dynamic environment variables to override any configuration option that appears in the agent section of the clearml.conf. For more information, see Dynamic Environment Variables.

note

ClearML's environment variables override the clearml.conf file, SDK, and configuration vault, but can be overridden by command-line arguments.

Name	Description
CLEARML_DOCKER_IMAGE	Sets the default docker image to use when running an agent in Docker mode
CLEARML_WORKER_NAME	Sets the Worker's name
CLEARML_WORKER_ID	Sets the Worker ID
CLEARML_CUDA_VERSION	Sets the CUDA version to be used
CLEARML_CUDNN_VERSION	Sets the CUDNN version to be used
CLEARML_CPU_ONLY	Force CPU only mode
CLEARML_DOCKER_SKIP_GPUS_FLAG	Skips the GPUs flag (support for docker V18)
CLEARML_AGENT_ABORT_CALLBACK_CMD	Path to a callback script to execute when a Task is aborted. The script will run for up to the maximum time set with `CLEARML_AGENT_ABORT_CALLBACK_TIMEOUT`. Set Task callback script using `-e CLEARML_AGENT_ABORT_CALLBACK_CMD=path/script.sh`.
CLEARML_AGENT_ABORT_CALLBACK_TIMEOUT	Maximum time (in seconds) the abort callback script (set with `CLEARML_AGENT_ABORT_CALLBACK_CMD`) is allowed to run. Set abort callback timeout using `-e CLEARML_AGENT_ABORT_CALLBACK_TIMEOUT=30`.
CLEARML_AGENT_DOCKER_ARGS_FILTERS	Set a whitelist of allowed Docker arguments. Only arguments matching the specified patterns can be used when running a task. Use `shlex.split` whitespace-separated format. For example: `CLEARML_AGENT_DOCKER_ARGS_FILTERS="^--env$ ^-e$"`
CLEARML_AGENT_DOCKER_ARGS_HIDE_ENV	Hide Docker environment variables containing secrets when printing out the Docker command. When printed, the variable values will be replaced by `********`. See `agent.hide_docker_command_env_vars`
CLEARML_AGENT_DISABLE_SSH_MOUNT	Disables the auto `.ssh` mount into the docker
CLEARML_AGENT_FORCE_CODE_DIR	Bypass repository cloning and/or applying any changes to the code. The agent will execute the specified entry point from the provided path.
CLEARML_AGENT_FORCE_UV	If set to `1`, force the agent to use UV as the package manager. Overrides the default manager set in the clearml.conf under `agent.package_manager.type`
CLEARML_AGENT_FORCE_EXEC_SCRIPT	Allows overriding the remote execution script to bypass repository cloning and execute code already available where the remote agent is running. Use `module:file.py` format to specify a module and a script to execute (e.g. `.:main.py` to run `main.py` from the working dir)
CLEARML_AGENT_FORCE_TASK_INIT	If set to `1`, ClearML Agent adds `Task.init()` to scripts that do not have the call, creating a Task to capture code execution information and output, which is then sent to the ClearML Server. If set to `0` and the script does not include `Task.init()`, the agent will capture only the output streams and console output, without tracking code execution details, metrics, or models.
CLEARML_AGENT_FORCE_SYSTEM_SITE_PACKAGES	If set to `1`, overrides default `agent.package_manager.system_site_packages: true` behavior when running tasks in containers (docker mode and k8s-glue)
CLEARML_AGENT_GIT_CLONE_VERBOSE	If set to `1`, `git clone` calls will report progress verbosely
CLEARML_AGENT_GIT_USER	Sets the Git user for ClearML Agent
CLEARML_AGENT_GIT_PASS	Sets the Git password for ClearML Agent
CLEARML_AGENT_GIT_HOST	Sets Git host (only sending login to this host)
CLEARML_AGENT_GIT_USE_MS_ENTRA_TOKEN	If set to `1`, enables authentication to Azure DevOps repositories using a Microsoft Entra token. The Azure token will be taken from the git password setting (`CLEARML_AGENT_GIT_PASS` or `agent.git_host`). `CLEARML_AGENT_GIT_USE_MS_ENTRA_TOKEN` and `CLEARML_AGENT_GIT_USE_AZURE_PAT` are mutually exclusive, only one can be set to `1`. For more information, see the Azure DevOps authentication guide.
CLEARML_AGENT_GIT_USE_AZURE_PAT	If set to `1`, enables authentication to Azure DevOps repositories using a Personal Access Token (PAT). The PAT will be taken from the git password setting (`CLEARML_AGENT_GIT_PASS` or `agent.git_host`). `CLEARML_AGENT_GIT_USE_MS_ENTRA_TOKEN` and `CLEARML_AGENT_GIT_USE_AZURE_PAT` are mutually exclusive, only one can be set to `1`. For more information, see the Azure DevOps authentication guide
CLEARML_AGENT_EXEC_USER	User for Agent executing tasks (root by default)
CLEARML_AGENT_EXTRA_DOCKER_ARGS	Overrides extra docker args configuration
CLEARML_AGENT_EXTRA_DOCKER_LABELS	List of labels to add to docker container. See Docker documentation.
CLEARML_EXTRA_PIP_INSTALL_FLAGS	List of additional flags to use when the agent installs packages. For example: `CLEARML_EXTRA_PIP_INSTALL_FLAGS=--use-deprecated=legacy-resolver` for a single flag or `CLEARML_EXTRA_PIP_INSTALL_FLAGS="--use-deprecated=legacy-resolver --no-warn-conflicts"` for multiple flags
CLEARML_AGENT_EXTRA_PYTHON_PATH	Sets extra Python path
CLEARML_AGENT_INITIAL_CONNECT_RETRY_OVERRIDE	Overrides initial server connection behavior (true by default), allows explicit number to specify number of connect retries)
CLEARML_AGENT_NO_UPDATE	Boolean. Set to `1` to skip agent update in the k8s pod container before the agent executes the task
CLEARML_AGENT_K8S_HOST_MOUNT / CLEARML_AGENT_DOCKER_HOST_MOUNT	Specifies Agent's mount point for Docker / K8s
CLEARML_AGENT_TEMP_STDOUT_FILE_DIR	Allows overriding the default `/tmp` location for agent temporary files
CLEARML_K8S_GLUE_START_AGENT_SCRIPT_PATH	Provide an alternate path to place the agent startup script generated inside a k8s task pod (instead of the default `~/~/__start_agent__.sh`)
CLEARML_AGENT_PACKAGE_PYTORCH_RESOLVE	Sets the PyTorch resolving mode. The options are: `none` - No resolving. Install PyTorch like any other package `pip` (default) - Sets extra index based on cuda and lets pip resolve `direct` - Resolve a direct link to the PyTorch wheel by parsing the pytorch.org pip repository, and matching the automatically detected cuda version with the required PyTorch wheel. If the exact cuda version is not found for the required PyTorch wheel, it will try a lower cuda version until a match is found
CLEARML_AGENT_DEBUG_INFO	Provide additional debug information for a specific context (currently only the `docker` value is supported)
CLEARML_AGENT_CHILD_AGENTS_COUNT_CMD	Provide an alternate bash command to list child agents while working in services mode
CLEARML_AGENT_SKIP_PIP_VENV_INSTALL	Instead of creating a new virtual environment inheriting from the system packages, use an existing virtual environment and install missing packages directly to it. Specify the Python binary of the existing virtual environment. For example: `CLEARML_AGENT_SKIP_PIP_VENV_INSTALL=/home/venv/bin/python`. Note: By default, ClearML Agent will still install any missing packages into the specified virtual environment. To ensure no additional packages will be installed, also set `CLEARML_AGENT_SKIP_PYTHON_ENV_INSTALL=1`.
CLEARML_AGENT_SKIP_PYTHON_ENV_INSTALL	If set to `1`, the agent will not install any required Python packages and will just use the preexisting Python environment to run the task.
CLEARML_AGENT_VENV_CACHE_PATH	Overrides venv cache folder configuration
CLEARML_MULTI_NODE_SINGLE_TASK	Control how multi-node resource monitoring is reported. The options are: `-1` - Only master node's (rank zero) console/resources are reported `1` - Graph per node i.e. machine/GPU graph for every node (console output prefixed with RANK) `2` - Series per node under a unified machine resource graph, graph per type of resource e.g. CPU/GPU utilization (console output prefixed with RANK)