From web-infra-dev-midscene-skills-1
Automates Android devices using Midscene.js from screenshots via ADB for taps, swipes, text input, app launches. Enables natural language control for mobile app testing, QA, and visual verification without DOM or accessibility labels.
npx claudepluginhub web-infra-dev/midscene-skillsThis skill is limited to using the following tools:
> **CRITICAL RULES — VIOLATIONS WILL BREAK THE WORKFLOW:**
Automates iOS devices via screenshots using Midscene CLI—no DOM or accessibility labels needed. Performs taps, swipes, typing, scrolling on any iOS app for testing and QA.
Automates Android, iOS, Aurora OS, and Desktop via CLI: screenshots, annotations, taps/swipes/text input, app install/launch/stop/uninstall, file push/pull, shell commands, device info queries.
Automates UI interactions on iOS simulators/devices and Android emulators/devices: open apps, snapshot/screenshot, tap/press, type/fill, scroll, extract UI info. For mobile QA, debugging, replay.
Share bugs, ideas, or general feedback.
CRITICAL RULES — VIOLATIONS WILL BREAK THE WORKFLOW:
- Never run midscene commands in the background. Each command must run synchronously so you can read its output (especially screenshots) before deciding the next action. Background execution breaks the screenshot-analyze-act loop.
- Run only one midscene command at a time. Wait for the previous command to finish, read the screenshot, then decide the next action. Never chain multiple commands together.
- Allow enough time for each command to complete. Midscene commands involve AI inference and screen interaction, which can take longer than typical shell commands. A typical command needs about 1 minute; complex
actcommands may need even longer.- Always report task results before finishing. After completing the automation task, you MUST proactively summarize the results to the user — including key data found, actions completed, screenshots taken, and any relevant findings. Never silently end after the last automation step; the user expects a complete response in a single interaction.
Automate Android devices using npx -y @midscene/android@1. Each CLI command maps directly to an MCP tool — you (the AI agent) act as the brain, deciding which actions to take based on screenshots.
act Can DoInside a single act call on Android, Midscene can tap, double-tap, long-press, type, clear text, scroll or swipe in any direction, pull to refresh, drag items, zoom with two fingers, press keys, and use system navigation such as Back, Home, or recent apps while working from the current visible screen.
Midscene requires models with strong visual grounding capabilities. The following environment variables must be configured — either as system environment variables or in a .env file in the current working directory (Midscene loads .env automatically):
MIDSCENE_MODEL_API_KEY="your-api-key"
MIDSCENE_MODEL_NAME="model-name"
MIDSCENE_MODEL_BASE_URL="https://..."
MIDSCENE_MODEL_FAMILY="family-identifier"
Example: Gemini (Gemini-3-Flash)
MIDSCENE_MODEL_API_KEY="your-google-api-key"
MIDSCENE_MODEL_NAME="gemini-3-flash"
MIDSCENE_MODEL_BASE_URL="https://generativelanguage.googleapis.com/v1beta/openai/"
MIDSCENE_MODEL_FAMILY="gemini"
Example: Qwen 3.5
MIDSCENE_MODEL_API_KEY="your-aliyun-api-key"
MIDSCENE_MODEL_NAME="qwen3.5-plus"
MIDSCENE_MODEL_BASE_URL="https://dashscope.aliyuncs.com/compatible-mode/v1"
MIDSCENE_MODEL_FAMILY="qwen3.5"
MIDSCENE_MODEL_REASONING_ENABLED="false"
# If using OpenRouter, set:
# MIDSCENE_MODEL_API_KEY="your-openrouter-api-key"
# MIDSCENE_MODEL_NAME="qwen/qwen3.5-plus"
# MIDSCENE_MODEL_BASE_URL="https://openrouter.ai/api/v1"
Example: Doubao Seed 2.0 Lite
MIDSCENE_MODEL_API_KEY="your-doubao-api-key"
MIDSCENE_MODEL_NAME="doubao-seed-2-0-lite"
MIDSCENE_MODEL_BASE_URL="https://ark.cn-beijing.volces.com/api/v3"
MIDSCENE_MODEL_FAMILY="doubao-seed"
Commonly used models: Doubao Seed 2.0 Lite, Qwen 3.5, Zhipu GLM-4.6V, Gemini-3-Pro, Gemini-3-Flash.
If the model is not configured, ask the user to set it up. See Model Configuration for supported providers.
npx -y @midscene/android@1 connect
npx -y @midscene/android@1 connect --deviceId emulator-5554
Use the dedicated launch step when you want a deterministic starting point before the rest of the task:
npx -y @midscene/android@1 launch --uri https://www.ebay.com
npx -y @midscene/android@1 launch --uri com.android.settings
npx -y @midscene/android@1 launch --uri com.android.settings/.Settings
Use this when the task needs lower-level device control that is not best expressed as a visible UI interaction:
npx -y @midscene/android@1 runadbshell --command "dumpsys battery"
This is forwarded to adb shell on the connected device. In practice, the underlying command is adb -s <deviceId> shell dumpsys battery and some environments may also include the default ADB server port, such as adb -P 5037 -s <deviceId> shell dumpsys battery.
npx -y @midscene/android@1 take_screenshot
After taking a screenshot, read the saved image file to understand the current screen state before deciding the next action.
Use act to interact with the device and get the result. It autonomously handles all UI interactions internally — tapping, typing, scrolling, swiping, waiting, and navigating — so you should give it complex, high-level tasks as a whole rather than breaking them into small steps. Describe what you want to do and the desired effect in natural language:
# specific instructions
npx -y @midscene/android@1 act --prompt "type hello world in the search field and press Enter"
npx -y @midscene/android@1 act --prompt "long press the message bubble and tap Delete in the popup menu"
# or target-driven instructions
npx -y @midscene/android@1 act --prompt "open Settings and navigate to Wi-Fi settings, tell me the connected network name"
When the user provides a screenshot, icon, logo, or reference image and wants an exact visual match, prefer tap --locate instead of a generic act --prompt. Pass --locate as JSON. The prompt describes the target, images supplies named reference images, and convertHttpImage2Base64: true is useful when the image URL may not be directly accessible to the model.
npx -y @midscene/android@1 tap --locate '{
"prompt": "tap the area contains the image",
"images": [
{
"name": "target image",
"url": "https://github.githubassets.com/assets/GitHub-Mark-ea2971cee799.png"
}
],
"convertHttpImage2Base64": true
}'
The same locate JSON shape also works for other commands that accept a locate parameter.
npx -y @midscene/android@1 disconnect
The generated HTML report is recommended for human reading first. It includes step-by-step execution details and replay videos for each operation, which makes it much easier to understand what happened and troubleshoot problems.
If another skill or tool needs to consume the report, first convert it with report-tool from the same platform CLI package. Prefer Markdown for LLM-based workflows. Use JSON when the report needs to be processed programmatically.
npx -y @midscene/android@1 report-tool --action to-markdown --htmlPath ./midscene_run/report/.../index.html --outputDir ./output-markdown
npx -y @midscene/android@1 report-tool --action split --htmlPath ./midscene_run/report/.../index.html --outputDir ./output-data
Since CLI commands are stateless between invocations, follow this pattern:
act to perform the desired action or target-driven instructions.adb shell am start -n <package/activity>) before invoking any midscene commands. Then take a screenshot to confirm the app is actually in the foreground. Only after visual confirmation should you proceed with UI automation using this skill. ADB commands are significantly faster than using midscene to navigate to and open apps."the Wi-Fi toggle switch on the right side" instead of "the toggle"."the search icon at the top right", "the third item in the list").act command: When performing consecutive operations within the same app, combine them into one act prompt instead of splitting them into separate commands. For example, "open Settings, tap Wi-Fi, and toggle it on" should be a single act call, not three. This reduces round-trips, avoids unnecessary screenshot-analyze cycles, and is significantly faster.tap --locate when a reference image is provided: If the user shares a screenshot, icon, or logo and wants that exact visual target, use tap --locate with a multimodal locate JSON object such as { "prompt": "...", "images": [...] } instead of relying only on act --prompt.Example — Popup menu interaction:
npx -y @midscene/android@1 act --prompt "long press the message bubble and tap Delete in the popup menu"
npx -y @midscene/android@1 take_screenshot
Example — Form interaction:
npx -y @midscene/android@1 act --prompt "fill in the username field with 'testuser' and the password field with 'pass123', then tap the Login button"
npx -y @midscene/android@1 take_screenshot
| Problem | Solution |
|---|---|
| ADB not found | Install Android SDK Platform Tools: brew install android-platform-tools (macOS) or download from developer.android.com. |
| Device not listed | Check USB connection, ensure USB debugging is enabled in Developer Options, and run adb devices. |
| Device shows "unauthorized" | Unlock the device and accept the USB debugging authorization prompt. Then run adb devices again. |
| Device shows "offline" | Disconnect and reconnect the USB cable. Run adb kill-server && adb start-server. |
| Command timeout | The device screen may be off or locked. Wake the device with adb shell input keyevent KEYCODE_WAKEUP and unlock it. |
| API key error | Check .env file contains MIDSCENE_MODEL_API_KEY=<your-key>. See Model Configuration. |
@midscene/* dependency version is outdated | Check local versions with npm ls @midscene/android @midscene/core @midscene/shared (or pnpm why @midscene/android). Compare with latest versions using npm view @midscene/android version, npm view @midscene/core version, and npm view @midscene/shared version. Upgrade as needed (npm i @midscene/android@latest @midscene/core@latest @midscene/shared@latest). |
| Wrong device targeted | If multiple devices are connected, use --deviceId <id> flag with the connect command. |