fnosp/fnos-dsh/harmony-automation

HarmonyOS Device Automation

CRITICAL RULES — VIOLATIONS WILL BREAK THE WORKFLOW:

1. 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.

2. 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.

3. 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 act commands may need even longer.

Automate HarmonyOS NEXT devices using npx -y @midscene/harmony@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.

What act Can Do

Inside a single act call on HarmonyOS, Midscene can tap, double-tap, long-press, type, clear text, scroll, drag items, press keys, and use system navigation such as Back, Home, or recent apps while working from the current visible screen. Two-finger zoom is not available because the underlying HarmonyOS automation layer does not expose multi-touch input.

Prerequisites

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.

HDC Setup

HDC (HarmonyOS Device Connector) must be installed and accessible. Common setup:

• Install via DevEco Studio
• Or set HDC_HOME environment variable to point to the HDC directory

Verify HDC is working:

hdc version
hdc list targets

Commands

Connect to Device

npx -y @midscene/harmony@1 connect
npx -y @midscene/harmony@1 connect --deviceId 0123456789ABCDEF

Launch an App or URL

Use the dedicated launch step when you want a deterministic starting point before the rest of the task:

npx -y @midscene/harmony@1 launch --uri com.huawei.hmos.settings
npx -y @midscene/harmony@1 launch --uri com.huawei.hmos.camera
npx -y @midscene/harmony@1 launch --uri https://www.example.com

Run a Raw HarmonyOS Shell Command

Use this when the task needs lower-level device control that is not best expressed as a visible UI interaction:

npx -y @midscene/harmony@1 runhdcshell --command "hidumper -s RenderService -a screen"

This is forwarded to hdc shell on the connected device. In practice, the underlying command is hdc -t <deviceId> shell hidumper -s RenderService -a screen.

Take Screenshot

npx -y @midscene/harmony@1 take_screenshot

After taking a screenshot, read the saved image file to understand the current screen state before deciding the next action.

Perform 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/harmony@1 act --prompt "type hello world in the search field and press Enter"
npx -y @midscene/harmony@1 act --prompt "long press the message bubble and tap Delete in the popup menu"
 
# or target-driven instructions
npx -y @midscene/harmony@1 act --prompt "open Settings and navigate to Wi-Fi settings, tell me the connected network name"

Assert Current Screen State

Use assert to verify that the current screen satisfies a natural language condition. It does not perform UI actions; it checks the visible screen state and passes only when the assertion is true. Use this for validation, QA checks, and final state verification after act.

npx -y @midscene/harmony@1 assert --prompt "there is a login button visible"
npx -y @midscene/harmony@1 assert --prompt "the settings screen shows Wi-Fi and Bluetooth options"
npx -y @midscene/harmony@1 assert --deviceId 0123456789ABCDEF --prompt "the app shows a successful login message"

By default a failed assertion throws an AI-generated reason. Pass --message to throw a custom error message instead, which is useful for surfacing the intended outcome in QA and CI logs.

npx -y @midscene/harmony@1 assert \
  --prompt "the order confirmation screen is visible" \
  --message "the order should be confirmed after tapping Pay"

When the assertion needs to compare against a reference image (icon, logo, screenshot), pass --image for the URL/path and --image-name for its display name. Each --image may be an http(s) link, a data: URI, or a local file path. Repeat both flags in matching order when you need to attach more than one image. Add --convertHttpImage2Base64 true when the model cannot reach the URL directly. Requires @midscene/harmony@1.9.0+.

npx -y @midscene/harmony@1 assert \
  --prompt "the visible app icon matches the supplied reference image" \
  --image "https://github.githubassets.com/assets/GitHub-Mark-ea2971cee799.png" \
  --image-name "icon" \
  --convertHttpImage2Base64 true
 
# or with a local file
npx -y @midscene/harmony@1 assert \
  --prompt "the header on screen matches the local screenshot" \
  --image "./fixtures/header.png" \
  --image-name "header"
 
# multiple reference images — pair --image and --image-name by order
npx -y @midscene/harmony@1 assert \
  --prompt "the screen shows both the app icon and the header" \
  --image "./fixtures/icon.png"   --image-name "icon" \
  --image "./fixtures/header.png" --image-name "header"

Record and Assert Transient UI

Use a recording when the state to verify may disappear before a current-screen assertion runs, such as a toast, loading banner, animation, or transition:

# Terminal 1: keep this foreground command running
npx -y @midscene/harmony@1 record start \
  --device-id 0123456789ABCDEF \
  --output ./submission-observation.json
 
# Terminal 2, while Terminal 1 records
npx -y @midscene/harmony@1 act --device-id 0123456789ABCDEF \
  --prompt "tap the Submit button"
 
# Send Ctrl+C to Terminal 1 and wait for the saved-path message, then assert
npx -y @midscene/harmony@1 assert --device-id 0123456789ABCDEF \
  --record ./submission-observation.json \
  --prompt "a success toast appeared during submission"

Pass target flags, capture flags, and --output to record start, then wait for Recording. Press Ctrl+C to stop and save. Keep the recorder as a foreground process in its dedicated terminal; never add shell &. Perform the interaction manually or from a second terminal, send Ctrl+C to the recorder, and wait until it prints the saved path before asserting. HarmonyOS recording uses periodic screenshots. Optional capture flags are --interval-ms, --max-frames, and --watchdog-ms; --max-frames caps sampled frames, and the manifest may contain one additional final representative frame. The default watchdog finalizes and saves the recording after five minutes, while --watchdog-ms 0 disables that safety limit. The output is a JSON manifest plus an adjacent <name>.frames image directory, not an encoded video or archive. The manifest contains relative JPEG/PNG paths and no base64 image bodies. Keep or move the JSON file and image directory together, and pass the JSON path to assert --record. Use ordinary assert without --record when only the current screen matters.

Use a Reference Image for Precise Targeting

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/harmony@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.

Disconnect

npx -y @midscene/harmony@1 disconnect

Consume Report Files

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/harmony@1 report-tool --action to-markdown --htmlPath ./midscene_run/report/.../index.html --outputDir ./output-markdown
npx -y @midscene/harmony@1 report-tool --action split --htmlPath ./midscene_run/report/.../index.html --outputDir ./output-data

Workflow Pattern

Since CLI commands are stateless between invocations, follow this pattern:

1. Connect to establish a session
2. Launch the target app and take screenshot to see the current state, make sure the app is launched and visible on the screen.
3. Execute action using act to perform the desired action or target-driven instructions. Use assert for the resulting screen state, or keep record start --output ... running in a dedicated terminal during transient-state workflows, stop it with Ctrl+C, and then use assert --record.
4. Disconnect when done

Best Practices