agentleFS
ruizricaruizrica/agent-pi/skills/qa-automation/qa-scroll/SKILL.md

qa-scroll

Skill274 starsChanged 8 months agoMITView on GitHub

QA test skill for verifying scroll-based media feeds — video autoplay, scroll navigation, mute/unmute, and playback progression. Uses the CDP + agent-device dual-driver architecture. Works with any React Native app using expo-video or similar video player libraries. Invoke when user says "test scroll feed", "test video playback", "QA the feed", "verify video autoplay", "run scroll tests", "test media player", or any task requiring scroll-based media feed verification.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Before running, set your app-specific values in qa.config.sh or qa.config.local.sh:

    line 45
  2. Distinctive:

    If your app exposes a scroll-to-next function via a debug hook:

    line 126
  3. Very specific:

    If no hook is available, the skill falls back to agent-device swipe gestures.

    line 139
  4. Very specific:

    Check the specific [SETUP] line that shows FAILED.

    line 161
  5. Distinctive:

    No video content loaded (API returned empty feed)

    line 177

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

qa-scroll

QA test skill for verifying scroll-based media feeds — video autoplay, scroll navigation, mute/unmute toggle, and playback progression. Uses the dual-driver architecture (CDP + agent-device) to test media feeds in native mobile apps.

What It Tests

TestMethodAssertion
First video autoplaysCDP player state queryplayer.playing === true, currentTime > 0
Video progress advancesCDP currentTime check after delaycurrentTime increased between checks
Scroll to next videoCDP scroll hook or agent-device swipeNew video starts playing
Mute toggleCDP player.muted = !player.mutedMute state flips
Unmute toggleCDP player.muted = !player.mutedMute state flips back
Scroll continuityMultiple swipesEach new video autoplays
Final route checkCDP cdp_get_routeStill on the feed screen

Setup Guard (Automatic)

The skill includes a setup guard that runs before every test. It checks and auto-fixes:

CheckWhat it does if missing
iOS Simulator bootedBoots it or auto-detects a booted one
Dev server runningStarts it in background, waits up to 60s
App in foregroundLaunches via xcrun simctl launch
CDP target availablePolls /json endpoint for up to 30s
CDP connection functionalSends eval 1+1, retries 3x
Navigation module ID validAuto-scans Metro modules with caching
Error overlaySuppresses LogBox errors via CDP

Configuration

Before running, set your app-specific values in qa.config.sh or qa.config.local.sh:

# Required
export APP_BUNDLE_ID="com.yourapp.dev"
export PROJECT_DIR="/path/to/your/project"

# Video player (for expo-video apps)
export VIDEO_PLAYER_CLASS="VideoPlayer"     # Class to look for
export GLOBAL_PLAYERS_VAR="__qaVideoPlayers" # Global tracking variable

# Screen names (for CDP navigation)
export SCREEN_EXPLORE="ExploreScreen"       # Your feed screen name

# Optional: if your app has a feed scroll hook
export GLOBAL_FEED_VAR="__qaFeedState"      # Feed state debug hook name

Test Result States

StatusMeaning
passedAssertion verified via CDP
failedAssertion verified but wrong
skippedCDP query inconclusive — cannot determine pass/fail

Architecture

CDP (Hermes Runtime)              agent-device (Simulator)
┌──────────────────────┐          ┌──────────────────────┐
│ navigate to tab      │          │ swipe up (scroll)     │
│ install debug hook   │          │ tap center (fallback) │
│ query player state   │          │ screenshot capture    │
│   .playing           │          │ appstate check        │
│   .muted             │          │                       │
│   .currentTime       │          │                       │
│ toggle mute via CDP  │          │                       │
│ dismiss error overlay│          │                       │
└──────────────────────┘          └──────────────────────┘

Why CDP, not agent-browser?

agent-browser uses Playwright's CDP protocol which sends Target.setDiscoverTargets — a method Hermes doesn't support. Native apps don't have a DOM to interact with. Raw CDP via WebSocket to Hermes is the correct approach for JS runtime queries in React Native apps.

Why agent-device, not agent-browser?

agent-device is purpose-built for iOS/Android simulator control — screenshots, swipe gestures, accessibility snapshots. agent-browser is for web pages. Native apps render native views, not web views.

Usage

Run the example test

bash .pi/skills/qa-automation/qa-scroll/run.sh

Run a specific flow

bash .pi/skills/qa-automation/qa-scroll/flows/example-scroll-test.sh

View results

  • Screenshots: /tmp/qa-tests/screenshots/<test-name>/
  • Report JSON: /tmp/qa-tests/<test-name>-report.json

Customizing for Your App

Step 1: Configure video player detection

The debug hook patches VideoPlayer.prototype.play() to track instances. If your app uses a different video player:

# In qa.config.sh:
export VIDEO_PLAYER_CLASS="MyVideoPlayer"  # Your player class name

The hook scans Metro modules looking for module.default.VideoPlayer or module.VideoPlayer. Adjust the scan in scroll-helpers.sh if your player is exported differently.

Step 2: Configure feed scrolling

If your app exposes a scroll-to-next function via a debug hook:

# In your app code (dev builds only):
globalThis.__qaFeedState = {
    currentIndex: 0,
    scrollToNext: () => { /* scroll logic */ },
    scrollToIndex: (i) => { /* scroll to index */ },
    getData: () => { /* return feed data array */ },
    getItem: (i) => { /* return item at index */ }
};

If no hook is available, the skill falls back to agent-device swipe gestures.

Step 3: Create your test flow

Copy flows/example-scroll-test.sh and customize the steps for your app's feed structure.

File Structure

qa-scroll/
├── SKILL.md                          # This file
├── lib/
│   ├── setup-guard.sh                # Prerequisites checker + auto-fixer
│   └── scroll-helpers.sh             # Video state, mute control, feed interaction
├── flows/
│   └── example-scroll-test.sh        # Example test (customize for your app)
└── run.sh                            # Runner with JSON report output

Troubleshooting

"Setup guard failed"

Check the specific [SETUP] line that shows FAILED. Common causes:

  • Simulator not installed or wrong UDID
  • Dev server port in use by another process
  • App not installed (build and install first)

"No CDP target found"

The app needs to be connected to the dev server. After a fresh install, launch the app and wait for the connection.

"VideoPlayer class not found"

The module scan didn't find your video player class. Check:

  • Is the video player library installed? (expo-video, react-native-video, etc.)
  • Does the class name match VIDEO_PLAYER_CLASS in config?
  • Try widening the scan range: export MODULE_SCAN_END=10000

Tests show "skipped"

CDP couldn't read player state. The debug hook may not have captured any players yet. Common causes:

  • No video content loaded (API returned empty feed)
  • Players created before hook was installed (hook captures on play() call)
  • Video library doesn't use the expected class structure

Error overlay appears

The setup guard suppresses LogBox, but errors during CDP eval may trigger new overlays. The test automatically checks for and dismisses overlays before screenshots.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.