skills/testworkflow-author/examples/command-args.yaml

# Example: Command/Args Pattern (Non-Shell Entrypoint)
# Patterns: command, args, entrypoint-based images
# What this demonstrates: Some container images are designed to be used
#                       with command/args (entrypoint-based) rather than
#                       shell. This shows both approaches side by side.

apiVersion: testworkflows.testkube.io/v1
kind: TestWorkflow
metadata:
  name: command-args
  labels:
    tool: postman
    category: api-testing
spec:
  content:
    git:
      uri: https://github.com/kubeshop/testkube
      revision: main
      paths:
      - test/postman/postman-executor-smoke.postman_collection.json

  container:
    resources:
      requests:
        cpu: 256m
        memory: 128Mi
    workingDir: /data/repo/test/postman

  steps:
  # Approach 1: shell — simplest, wraps everything in sh -c.
  # Use when the image has a default entrypoint and you just need
  # to pass arguments to it.
  - name: Run with shell
    run:
      image: postman/newman:6-alpine
      shell: newman run postman-executor-smoke.postman_collection.json --env-var TESTKUBE_POSTMAN_PARAM=TESTKUBE_POSTMAN_PARAM_value

  # Approach 2: args — directly passes arguments to the image's
  # ENTRYPOINT without a shell wrapper. Use when the image's
  # entrypoint is the command itself (e.g., newman, k6, artillery).
  # This is more efficient (no sh -c overhead) and avoids shell
  # escaping issues.
  - name: Run with args
    run:
      image: postman/newman:6-alpine
      args:
      - run
      - postman-executor-smoke.postman_collection.json
      - --env-var
      - TESTKUBE_POSTMAN_PARAM=TESTKUBE_POSTMAN_PARAM_value