Connect to Anaconda Platform and run your first flow

There are two main ways to develop and execute code on Anaconda Platform:

• You can use cloud workstations through VS Code, Cursor, Windsurf, or SSH, backed by cloud instances hosted as part of the platform in your cloud account.
• You can access the platform from any existing development environment, such as a laptop or a cloud-based environment.

​

Get access to the platform

Admin step

Before you can authenticate, an administrator needs to invite you to the platform by adding your email in the user management view. Once you are invited, you can log in to the platform through your configured SSO provider.

Tokens are personal. Users should not share their personal access tokens. If you need shareable access tokens that are not tied to a person, see Programmatic access via machine users.

​

Connect to the platform

You can connect to the platform from your existing development environment or with a cloud workstation.

If you have an existing installation of open-source Metaflow in your environment and you want to use Anaconda Platform in that same environment, follow instructions for Use my existing Metaflow environment.

Use my existing environment

1. Click your profile in the lower-left corner to open your Local Setup page:

   

2. Open a terminal and install the outerbounds package:

   conda

   conda install outerbounds
   

   uv

   uv pip install --upgrade outerbounds
   

   pip

   pip install -U outerbounds
   
   You can install the package in an isolated environment, but it is not required.

3. Copy the configuration command shown on your Local Setup page and run it in your terminal:

   outerbounds configure <CONFIGURATION_STRING>
   

   The configuration command saves your access token in a system-wide configuration file. You are now ready to start using the platform.

   If you encounter problems connecting to the platform, run outerbounds check -v to validate your setup. All checks should display ...OK. If your checks do not pass, contact your administrator for help.

Use a cloud workstation

You can develop with cloud workstations by following these steps:

1. If necessary, add a workstation for yourself in the platform.
2. To start your cloud workstation, turn on the toggle beside the Actions dropdown. It might take a few minutes for the workstation to start.
3. Open the Actions dropdown and choose to connect to the workstation either through the Web UI, or in VSCode Desktop.

If you encounter problems connecting to your workstation from VSCode, run outerbounds check -w in the terminal to validate your setup. All checks should display ✅. If your checks do not pass, contact your administrator for help.

Use my existing Metaflow environment

1. Click your profile in the lower-left corner to open your Local Setup page:

   

2. Create a new environment for the outerbounds package to avoid conflicts between metaflow and outerbounds:

   conda

   conda create --name ob python=3.12 outerbounds
   conda activate ob
   

   venv

   python -m venv ob
   source ob/bin/activate
   conda install outerbounds
   

3. When configuring the access token, use the --profile option to create a new configuration profile instead of overwriting your existing Metaflow config:

   outerbounds configure --profile ob <CONFIGURATION_STRING>
   

4. To use the platform, set an environment variable:

   export METAFLOW_PROFILE=ob
   

If you encounter problems connecting to the platform, run METAFLOW_PROFILE=ob outerbounds check -v in the terminal to validate your setup. All checks should display ...OK. If your checks do not pass, contact your administrator for help.

​

Run your first flows

Once you have configured access to the platform, you can run a simple Metaflow flow to confirm that everything works.

If you are new to Metaflow, take a quick look at the Metaflow documentation. All features of Metaflow work on Anaconda Platform.

​

Hello world

1. In your IDE, create a directory named first-flow with a file inside called hello.py, or create the directory and file by running the following commands in your terminal:

   mkdir first-flow && cd first-flow
   touch hello.py
   
   This command creates a new directory inside your current working directory.

2. Save the following code snippet in hello.py:

   from metaflow import FlowSpec, step
   
   class HelloFlow(FlowSpec):
   
       @step
       def start(self):
           print("Hello world! 👋")
           self.next(self.end)
   
       @step
       def end(self):
           pass
   
   if __name__ == "__main__":
       HelloFlow()
   

3. Run the flow:

   python hello.py run
   

   If you see “Hello world! 👋” in the terminal’s return, the flow ran successfully. You should also see a link to the Runs view, which shows all workflow executions taking place on the platform.

   

​

Hello artifacts

Metaflow automatically saves any data you assign to a self. variable as an artifact. Artifacts let your steps share data and let you inspect that data after a run finishes.

1. Replace the contents of hello.py with the following code snippet:

   from metaflow import FlowSpec, step, card
   
   class HelloFlow(FlowSpec):
   
       @card
       @step
       def start(self):
           self.greeting = "Hello world!"
           self.x = 1
           self.next(self.end)
   
       @card
       @step
       def end(self):
           print(self.greeting)
           self.x += 10
           print("x is", self.x)
   
   
   if __name__ == "__main__":
       HelloFlow()
   

   The start step assigns values to self.greeting and self.x, and the end step reads and updates them. Metaflow moves this data between steps for you, even when the steps run on separate cloud instances, as they do when you scale your flows to the cloud.

2. Run the flow again:

   python hello.py run
   

3. Open the Runs view and select the latest HelloFlow run.

4. Select the start task, then select the Cards tab:

   

   The @card decorator adds this card to the task view. To see how the value of x changes between steps, compare the cards for the start and end tasks.

Metaflow versions artifacts automatically, so you can track metrics, models, dataframes, or any other data across prototype and production runs. To learn what else cards can display, see Visualizing results in the Metaflow documentation.