Feat: Adds interactive session creation and listing

[l-mansouri]

Overview

This PR implements the first two commands of the interactive session module of the CLI: list and create.

JIRA

Please add here as many related tasks this PR covers with its brief description, if more than one ticket

• LP-105728
• LP-105729
  -LP-106300

Changes

• Implements the interactive-session module
• Implements the listing of interactive session
• Implements the creation of interactive session
• Refactors the documentation to add the new implemented module and commands

Acceptance Criteria

This acceptance criteria is tested in ADAPT workspace in PROD using the default profile

LISTING

Scenario 1: Successfully list all IA sessions in the workspace

cloudos interactive-session list

Scenario 2: Listed sessions can report the same information as the platform UI

cloudos interactive-session list --table-columns status,name,owner,project,created_at,runtime,saved_at,cost,resources,backend,version

Scenario 3: User can retrieve a specific page of results by providing page number and page-size

cloudos interactive-session list --page 3 --limit 5

Scenario 4: User can navigate through pages interactively using the terminal

cloudos interactive-session list

Screen.Recording.2026-03-18.at.12.13.37.mov

Scenario 5: Attempt to list IA sessions with invalid credentials

cloudos interactive-session list --apikey invalid_key 

Scenario 6: User can filter sessions by status

# Filter by setup status
cloudos interactive-session list --filter-status setup

# Filter by initialising status
cloudos interactive-session list --filter-status initialising

# Filter by running status
cloudos interactive-session list --filter-status running

# Filter by stopped status
cloudos interactive-session list --filter-status stopped

Scenario 7: User can filter to display only their own sessions

cloudos interactive-session list --filter-owner-only

Scenario 8: User can view archived sessions

cloudos interactive-session list --archived

Scenario 9: User can export session list in CSV format

# Export with default filename
cloudos interactive-session list --output-format csv

# Export with custom filename and all fields
cloudos interactive-session list --output-format csv --all-fields --output-basename my_sessions

Scenario 10: User can export session list in JSON format

cloudos interactive-session list --output-format json

Scenario 11: User can select specific columns to display in the table output

cloudos interactive-session list --table-columns id,name,status,type,instance,cost,owner

AZURE LISTING

cloudos interactive-session list --profile azure-prod

CREATING

Scenario 1: Successfully create a new Jupyter IA session

cloudos interactive-session create --session-type jupyter --name test_jupyter

Scenario 2: Successfully create a new VSCode IA session

cloudos interactive-session create --session-type vscode --name test_vs

Scenario 3: Successfully create a new RStudio IA session

cloudos interactive-session create --session-type rstudio --name test_rstudio

Scenario 4: Successfully create a new Spark IA session

cloudos interactive-session create --session-type spark --name test_spark

Scenario 5: Create a new IA session with custom instance type

cloudos interactive-session create --session-type jupyter --name test_instance --instance c5.2xlarge

Scenario 6: Create a new IA session with custom storage size

cloudos interactive-session create --session-type jupyter --name test_storage --storage 1000

Scenario 7: Create a new IA session with spot instance flag

cloudos interactive-session create --session-type jupyter --name test_spot --spot

Scenario 8: Create a new IA session with shutdown timeout

cloudos interactive-session create --session-type jupyter --name test_time --shutdown-in 10m

Scenario 9: Create a new IA session with cost limit

cloudos interactive-session create --session-type jupyter --name test_cost --cost-limit 0.05

Scenario 10: Create a new IA session with shared flag

cloudos interactive-session create --session-type jupyter --name test_public --shared

Scenario 11: Create a new RStudio IA session with specific R version

cloudos interactive-session create --session-type rstudio --name test_rstudio --r-version 4.4.2

Scenario 12: Create a new Spark IA session with custom master and worker configuration

cloudos interactive-session create --session-type spark --name test_spark --spark-master c5.xlarge --spark-workers 2 --spark-core c5.xlarge

Scenario 13: Create a new IA session with linked file explorer data

cloudos interactive-session create --session-type jupyter --name test_link_fe --link leila-test/AnalysesResults/JG_1shard_chr15-68f210f9e2fdcb612f8e6fe8/results/pipeline_info

Scenario 14: Create a new IA session with linked S3 data

cloudos interactive-session create --session-type jupyter --name test_link_s3 --link s3://lifebit-featured-datasets/pipelines/phewas/example-data/

Scenario 15: Create a new IA session with mounted S3 data

cloudos interactive-session create --session-type jupyter --name test_mount_s3 --mount s3://lifebit-featured-datasets/pipelines/phewas/100_binary_pheno.phe

Scenario 16: Create a new IA session with mounted file explorer data

cloudos interactive-session create --session-type jupyter --name test_mount_fe --mount leila-test/Data/benchmark_test.txt

<img width="1040" height="498" alt="Screenshot 2026-03-18 at 15 26 10" src="https://github.com/user-attachments/assets/48b4cd81-1c82

-4642-a431-ec153c092605" />

Scenario 17: Attempt to create a session with an unsupported session type

cloudos interactive-session create --session-type invalid_type --name test_invalid

Scenario 18: Attempt to create a session with invalid credentials

cloudos interactive-session create --session-type jupyter --name test_auth --apikey invalid_key --cloudos-url https://test.com

Scenario 19: Attempt to create a session with missing required parameters

cloudos interactive-session create --session-type jupyter

AZURE IA CREATION

cloudos interactive-session create --profile prod-azure --name Azure_test --session-type rstudio --r-version 4.5.2 --project-name Test-cli

[danielboloc]

thanks @l-mansouri

I have tested it in my local env, and I have some questions:

1. Using --table-columns with --help it just displays an empty table, but if I use --table-columns id for example, it does show only the help. Could you please check what could happen?

2. In Scenario 2 of listing you are using cloudos interactive-session list --table-columns status,name,owner,project,created_at,runtime,saved_at,cost,resources,backend,version, but these columns do not appear as options in the help section

3. Could we change the Type of the sessions to not appear as awsRstudio,awsVSCode,awsJupyterNotebook raw values, and have them as in the Platform UI?

4. Could we change --filter-owner-only to --filter-only-mine as in the job list subcommand? Or we could do the reverse, change the jobs one as this one, but let's have them unified in the option name

5. --output-format csv does not show user name, maybe because it uses user.name?

6. Why the json output format is limited to 10 sessions, but not csv? Why this limit?

7. Should we do a check for instance type? I have added a non-existent type and it send the session

8. Should we also add a new column for spot type instances, either when creating the session or when listing it?

9. The same as for --shutdown-in (how much time has left), --cost-limit (how much cost is left), --shared

thank you

[dapineyro]

Hey @l-mansouri ! Huge work here, that's why there are so many comments. But it looks great overall. Thanks!

[dapineyro]

Suggested change

	cloudos interactive-session create --cloudos-url $CLOUDOS_URL --apikey $CLOUDOS_TOKEN --workspace-id $CLOUDOS_WORKSPACE_ID --project-name "$PROJECT_NAME" --name ci_test_cli --session-type jupyter --shutdown-in 10m
	cloudos interactive-session create --cloudos-url $CLOUDOS_URL --apikey $CLOUDOS_TOKEN --workspace-id $CLOUDOS_WORKSPACE_ID --project-name "$PROJECT_NAME" --name ci_test_cli --session-type jupyter --shutdown-in 1m

Why to keep it open 10m?

[l-mansouri]

because in 1 minute the session is not fully running. And also for the rest of the commands that will require this one to work

[dapineyro]

Suggested change

	cloudos interactive-session create --cloudos-url $CLOUDOS_URL --apikey $CLOUDOS_TOKEN --workspace-id $CLOUDOS_WORKSPACE_ID --project-name "$PROJECT_NAME" --name ci_test_cli --session-type jupyter --shutdown-in 10m --execution-platform azure --instance $INSTANCE_TYPE
	cloudos interactive-session create --cloudos-url $CLOUDOS_URL --apikey $CLOUDOS_TOKEN --workspace-id $CLOUDOS_WORKSPACE_ID --project-name "$PROJECT_NAME" --name ci_test_cli --session-type jupyter --shutdown-in 1m --execution-platform azure --instance $INSTANCE_TYPE

[dapineyro]

        cloudos interactive-session create --cloudos-url $CLOUDOS_URL --apikey $CLOUDOS_TOKEN --workspace-id $CLOUDOS_WORKSPACE_ID --project-name "$PROJECT_NAME" --name ci_test_cli --session-type jupyter --shutdown-in 1m

[l-mansouri]

Thanks Dani and David, I should have addressed all your comments.

@danielboloc regarding your questions:

1. It was a misinterpretation of the flags. I have fixed that by implementing a parsing and check of the column names against the list of available columns.

2. the list of available columns is now added to the help text 3. you're right, it makes the table more human readable. I have implemented a mapping function

4. the flag now is --filter-only-mine as the other commands.

5. I have fixed that there was a wrong mapping on my side. Thank you for spotting it
6. Both csv and json were already returning 10 results (the number is changeable with the --limit flag). Now both of the execution print a message with the number of lines. The only reason why the two files output only 10 values is to not create huge files that can be hard to navigate, but what is outputted is fully customisable using the --limit --page flags

7. I have added a function that prevents typos, but your test would still go through, even if the session is not created as the endpoint does not return an error and the regex for the check must be flexible enough to cover a variety of potential instances. (the regex is looking for this pattern aws_pattern = r'^[a-z]{1,4}\d+[a-z]*\.(\d+)?(nano|micro|small|medium|large|xlarge|metal)$' and azure_pattern = r'^(Standard|Basic)_[A-Z]\d+[a-z]*(_v\d+)?$')

8. e 9. I have added specific columns to the listing to report on all those things.

@dapineyro Thanks for spotting some of the left-overs of not-working implemenations! I have reviewed and tidied the README.