agentsclimarketplace

Run test case

Skill hoangdh2001/apple-app-ui-test-automation/skills/run-test-case

Execute a single UI test case and return its result. Works across iPhone, iPad, and macOS — pass the platform param to adjust behavior. Requires the app to already be built and running before calling this skill. Returns: pass, flaky, or fail — with screenshot path, logs, and root cause on failure.From its SKILL.md

Install
npx -y skills add hoangdh2001/apple-app-ui-test-automation --skill run-test-case

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

3.5 KB, as published. Nobody here has run it

Skill: run-test-case

Input

  • test_case — one parsed test case object (id, name, feature, steps, expect)
  • platform — "iphone" | "ipad" | "macos"
  • udid — simulator UDID (omit for macOS)
  • bundle_id — app bundle identifier
  • report_dir — path to write screenshots, e.g. tests/reports/screenshots

Step 1: Reset app state

iPhone / iPad:

Navigate to a known base state before running:

axe tap --label Dashboard --udid <UDID> --wait-timeout 3

If this fails (modal blocking, crash):

xcodebuildmcp simulator stop --simulator-id <UDID> --bundle-id <bundle>
xcodebuildmcp simulator launch-app --simulator-id <UDID> --bundle-id <bundle>

Then poll until interactive (max 30s, retry every 3s):

axe describe-ui --udid <UDID>
# Success: root element has "type": "Application"

macOS: Skip reset. macOS apps do not have a tab bar to navigate to. Proceed directly to Step 2.


Step 2: Execute interactions

Translate all YAML steps into a single axe batch call via stdin.

iPhone / iPad:

axe batch --udid <UDID> --stdin << 'EOF'
tap --label <label> --wait-timeout 5
type '<text>'
EOF

macOS:

axe batch --stdin << 'EOF'
tap --label <label> --wait-timeout 5
type '<text>'
EOF

Rules:

  • Always use --stdin, not --step or --file
  • Add --wait-timeout to every tap and navigate step

Step 3: Verify expected state

Run describe-ui in a separate call (batch is fire-and-forget):

iPhone / iPad: axe describe-ui --udid <UDID> macOS: axe describe-ui

For each expect entry:

  • ui_contains: "X" → assert element with AXLabel == "X" exists
  • element_not_exists: "X" → assert NO element with AXLabel == "X" exists

If all pass, capture screenshot and return ✅:

iPhone / iPad: axe screenshot --udid <UDID> --output <report_dir>/<id>-pass.png macOS: axe screenshot --output <report_dir>/<id>-pass.png


Step 4: Handle failure

If any assertion fails:

  1. Capture error screenshot:

    • iPhone/iPad: axe screenshot --udid <UDID> --output <report_dir>/<id>-error.png
    • macOS: axe screenshot --output <report_dir>/<id>-error.png
  2. Start log capture:

    xcodebuildmcp logging start-simulator-log-capture \
      --simulator-id <UDID> \       # omit on macOS
      --bundle-id <bundle> \
      --capture-console true
    # Save the returned logSessionId
    
  3. Retry the failing step only with --wait-timeout 5, then re-run describe-ui and assert again.

  4. Stop log capture:

    xcodebuildmcp logging stop-simulator-log-capture --log-session-id <logSessionId>
    
  5. Determine result:

    • Retry passed → ⚠️ flaky
    • Retry failed → ❌ fail
  6. Suggest root cause from logs:

    • "Element not found" → label changed, wrong screen, animation still running
    • "Multiple matches" → use coordinate-based tap fallback
    • Crash in logs → report crash type

Output

{
  id: string
  name: string
  feature: string
  result: "pass" | "flaky" | "fail"
  duration_s: number
  screenshot: string
  error?: string
  logs?: string
  suggested_cause?: string
}

Keep looking

Skills are one crate of 609,689. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.