UiPath Robot

The UiPath Robot Guide

Linux Robots


Using Robots in Linux environments allows you to run unattended automations in Docker containers and provides a faster experience and an easier way of scaling up your deployment.

When Linux environments are used, robots can only run processes developed as cross-platform in Studio that do not require UI Interaction.


Developing Cross-Platform Automations

Linux robots can only run unattended automations. When creating an automation, make sure to select the Cross-platform compatibility. This way, only cross-platform activities are shown, while unsupported ones are hidden.

The default activities packages for cross-platform projects are:

  • UiPath.System.Activities
  • UiPath.WebAPI.Activities
  • UiPath.UIAutomation.Activities


In order to deploy unattended robots in a Linux environment, follow the steps below:

Orchestrator Configuration

Create an unattended robot in orchestrator, as shown in this document, then save the machine key so it can be used later.

Client Machine

  1. Open a command line terminal and get the docker image using the following command:
docker pull
docker pull<tag>
  1. Start the downloaded image using the parameters to connect the Robot to Orchestrator using the command below:
docker run -e LICENSE_AGREEMENT=accept -e ORCHESTRATOR_URL="" -e MACHINE_KEY="$KEY" -tid
docker run -e LICENSE_AGREEMENT=accept -e ORCHESTRATOR_URL="" -e CLIENT_ID="$Client_ID" -e CLIENT_SECRET="$Client_secret" -tid

:information-source: To retrieve or start a specific version of the robot, replace the <tag> parameter with the actual version from the Docker Image Tags column in the table below:

Robot Version

Docker Image Tags



Not using any tag retrieves the most recent enterprise patch for the latest enterprise version.


Troubleshooting tip:

If you encounter issues connecting the robot to Orchestrator, this might be caused by a host machine configuration (e.g. firewall rules). To resolve this, add a specific DNS server in the command.
Example: docker run --dns="" -e LICENSE_AGREEMENT=accept -e ORCHESTRATOR_URL="" -e MACHINE_KEY="$KEY" -tid
Where refers to the public DNS resolver from Cloudflare. This can be any DNS resolver that can resolve the SignalR Hub.



In order to successfully start the robot, you must accept the license agreement by adding the -e LICENSE_AGREEMENT=accept parameter in the command.

Every time you run the command presented in step 2, a new robot is spawned in Orchestrator.

  1. To see the status of your robots on the machine use the following command:
docker ps -a
  1. To disconnect a robot use the command below:
docker stop {container_id}



Using docker kill container_id stops the robot on the machine but does not disconnect it from Orchestrator.

  1. To reconnect a specific robot use the command below:
docker start {container_id}


If you need to persist log files beyond the lifecycle of the container instance, use the following command:

docker run -e LICENSE_AGREEMENT=accept -e MACHINE_KEY="{machine_key}" -e ORCHESTRATOR_URL="" -v C:\Users\\Desktop\logs:/root/.local/share/UiPath/Logs/ -ti



The Robot Linux image is based on the Alpine Linux version of the All files required by the application are placed in the /root/application directory.

Running Jobs on Linux Robots

The procedure for running an unattended job on Linux robots is identical to the one for Windows.

Remote Debugging

To test your process, you can enable debug mode using the following command:

docker run -e LICENSE_AGREEMENT=accept -e DEBUG_SESSION=true -ti -p8573:8573

The command starts a new Robot Container with the Debug feature enabled running on port 8573. After the container is started, you can connect to the docker image using the host machine ip and the port 8573 making use of the Studio Remote Debugging feature.

If you want to debug an already started Robocontainer, you can use the unattended roboot connection feature with a robot container connected to the orchestrator.

Updated about a month ago

Linux Robots

Suggested Edits are limited on API Reference Pages

You can only suggest edits to Markdown body content, but not to the API spec.