본문으로 건너뛰기
cccskills
무료GitHub에 공개

developer-device-platform-basics

Provides guidance and instructions on managing remote devices on Developer Device Platform (DDP). Use when reserving remote Android devices, establishing connection tunnels, checking session status, or extending/cancelling leases. Don't use for iOS or local device/hardware inquiries.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

연동 도구Google Cloud
설치 방법 보기

포함된 파일(11)

  • SKILL.md9.8 KB
  • references/cancel_session.md314 B
  • references/describe_device.md388 B
  • references/list_sessions.md454 B
  • references/reserve_device.md663 B
  • references/session_status.md217 B
  • references/start_adb_forwarder.md2.2 KB
  • references/update_session_expiration.md439 B
  • references/view_device.md377 B
  • scripts/demo_adb_forwarder.py12.1 KB
  • scripts/requirements.txt71 B

SKILL.md(원문)

설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.

Developer Device Platform

Developer Device Platform (DDP) is a Google fully managed, global infrastructure providing access to a wide variety of physical and virtual devices.

[!WARNING] Developer Device Platform (DDP) is currently at Preview.

[!IMPORTANT] For all devicerun and devicestreaming API operations (reserving, status checking, stopping/canceling, updating, or listing a session), always verify and use the exact instructions and curl commands provided in the linked reference .md files.

Authentication & Setup

CRITICAL: Before running any requests, you MUST ensure the environment is correctly initialized by following these steps:

Before running any requests, verify if the gcloud executable is present. If missing, refer to the official Google Cloud CLI Installation Guide to install it on the current platform (Linux, macOS, Windows, etc.).

  1. Google Cloud Authentication: Authenticate with your Google Cloud credentials and configure active Application Default Credentials (ADC) for the Developer Device Platform:

    gcloud auth login --no-browser
    gcloud auth application-default login --no-browser
    
  2. Enable APIs (if not already enabled):

    gcloud services enable devicerun.googleapis.com devicestreaming.googleapis.com testing.googleapis.com --quiet
    

[!NOTE] Cloud Testing API is needed for Device Streaming API during Preview.

  1. Enable gcloud beta component:

    gcloud components install beta
    
  2. Setup Environment Variables: Set up the required project variable and access token:

    export PROJECT_ID=$(gcloud config get project)
    export ACCESS_TOKEN=$(gcloud auth application-default print-access-token 2>/dev/null)
    
  3. Python Environment: For instructions on setting up the python virtual environment, see start_adb_forwarder.md.

Listing Available Devices

To find the correct modelCode and osVersion to use when starting a session, you can list the available devices:

  1. List Models: Run the following command to list available Android device models:

    gcloud beta device-run devices list
    

    Use the ID column to find the value for the CATALOG_ID parameter to describe a specific device (e.g., shiba-36).

  2. Describe a Model: Run the API request to get more details about a specific model (e.g., supportedProducts, resolution). Always rely on the exact curl command and instructions provided in describe_device.md.

Starting a Device Session

When the user asks to reserve or connect to a device:

  1. Check Device Availability:

    Look up CATALOG_ID of the device using Listing Available Devices instructions. Check the device availability of the CATALOG_ID by using the exact curl command and instructions provided in describe_device.md. The device MUST contain "deviceStreaming" in "supportedProducts" to be reserved.

    If no specific device is specified, use CATALOG_ID=shiba-34 (Pixel 8 on SDK 34).

    If OS_VERSION was not specified by the user, ask the user to select a version from the device list (preferring the version with the highest availability {"available": "AVAILABILITY_HIGH" }).

    If OS_VERSION is unavailable for deviceStreaming, do not reserve one. Prompt the user for an alternative OS_VERSION.

  2. Extract Parameters:

    • model_id: modelCode from device details. Required.
    • version_id: osVersion from device details. Required.
  3. Reserve Device:

    Rule: Explicit User Confirmation Required. Reserving a device incurs billing charges and creates cloud resources. The agent MUST ALWAYS warn the user explicitly about the billing costs that will be incurred on the active Google Cloud project (e.g., ${PROJECT_ID}). You MUST STOP and ask for explicit approval before proceeding with any session creation commands.

    Then, run the API request with model_id and version_id to reserve the device. Always rely on the exact curl command and instructions provided in reserve_device.md.

    Parse the response to get session_name (the session name, e.g., projects/${PROJECT_ID}/deviceSessions/session-xxxxxx). If reservation fails, report the error.

  4. Wait for Session to be Active:

    While waiting for the device session to be provisioned, poll the session status until "state" is "ACTIVE". See session_status.md for the exact curl command.

    Repeat this check every 5 seconds to prevent hitting API rate limit. If it does not become active within 2 minutes (typically under 1 minute), report failure and cancel the session. Once active, extract expireTime from the session JSON response and convert it to the user's local time in a human-readable format (e.g., "June 9, 2026 at 2:44 PM PDT").

  5. Start Connection Forwarder: Start the ADB forwarder script to forward connection to the remote device. Always rely on the exact command and instructions provided in start_adb_forwarder.md. Ensure you record the Command ID.

  6. Wait for Online and Parse Port: Wait for the forwarder to be online and extract the listening port. Always rely on the exact logic and instructions provided in start_adb_forwarder.md.

  7. Provide Instructions to User:

    Once online, run adb -s localhost:{port} shell getprop ro.product.model to retrieve the device model name. Then, print a message directly to the user in the chat (do NOT create any artifact file) with the following instructions:

    Device is ready!

    Device Model: {device_model}
    OS Version: {version_id}
    ADB Address: localhost:{port}
    Session Expiration: {expire_time_human_readable_local}
    
  8. Save Session State: Save the {session_name} and {command_id} in your conversation memory/context so you can clean it up later.

Viewing the Device Screen of a Reserved Device

The coding agent can directly interact with the remote device using adb. Users may use a utility to display the screen and manually control the reserved device in DDP. See view_device.md for an example utility.

Stopping a Device Session

When the user asks to stop, cleanup, or release the device:

  1. Identify Session: Retrieve the active {session_name} and {command_id} from your context. If you don't have them, list active sessions first (see helper command below) to find the session name.

  2. Cancel Session via API: Cancel the session via the API. Always rely on the exact curl command and instructions provided in cancel_session.md.

  3. Terminate Connection Forwarder: Terminate the background process matching {command_id} using your environment's process management capability.

  4. Confirm: Confirm to the user that the session has been cancelled and resources released.

Change Device Session Expiration Time

When the user asks to change the expiration time of an active device session:

Rule: Explicit User Confirmation Required. Extending a device session incurs additional billing charges and creates cloud resources. The agent MUST ALWAYS warn the user explicitly about the extra billing costs that will be incurred on the active Google Cloud project (e.g., ${PROJECT_ID}). You MUST STOP and ask for explicit approval before proceeding with any session extension commands.

  1. Extract Parameters:

    • session_name: The active session name.
    • ttl: The new remaining duration (e.g., 3600s). Derive the ttl if it's provided in another format.
  2. Change Session via API: Change the session via the API using updateMask=ttl. Always rely on the exact curl commands and instructions provided in update_session_expiration.md.

  3. Restart Connection Forwarder:

    • Run adb disconnect localhost:{port} to ensure the old forwarder connection is closed.
    • Stop the old connection forwarder corresponding to {command_id}.
    • Start a new connection forwarder by following Step 5 in "Starting a Device Session" (calculating the new --ttl duration in seconds and storing the newly returned Command ID).
  4. Confirm: Confirm to the user that the session duration has been updated and the connection forwarder has been restarted with the new TTL.

Helper: List Active Sessions

To find active sessions if you lost context, always rely on the curl command and instructions provided in list_sessions.md.

References

리뷰

아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.

댓글

로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.

아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.

보내면 바로 공개됩니다. 이메일 주소나 전화번호 같은 개인 정보는 쓰지 마세요. URL은 2개까지입니다.

0 / 1,000자

이 양식은 reCAPTCHA로 보호되며 Google 개인정보처리방침 및 서비스 약관이 적용됩니다.

같은 저장소의 스킬

Configures best-practice alerting policies for AI agents using OpenTelemetry (OTel) metrics, generating output as Terraform (.tf) configuration files. Use when analyzing, writing, or deploying alerting policies to monitor agent latency, error rates, token usage, and quality metrics. Don't use for standard infrastructure monitoring unrelated to AI agents, or when the agent is not instrumented with OpenTelemetry (for Reliability, Cost, Safety, Security alerts). NOTE: Reliability, Cost, Safety, and Security alerts use generic OTel metrics and work across runtimes (such as Cloud Run, Vertex AI). Quality alerts rely on Vertex AI Online Monitors and are strictly bound to Vertex AI deployments.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

vaila-multimodaltoolbox/vaila212026년 10월 19일 업데이트

Deploy open models or custom weights from Model Garden to Agent Platform endpoints, check the status of an in-progress deployment operation, or clean up resources by undeploying models and deleting endpoints. Use when asked to actively deploy a model, list the Model Garden CATALOG of available models, check if a specific model is deployable (`gcloud ai model-garden models list-deployment-config`), query deployment cost, troubleshoot deployment errors (like quota limits), or undeploy/clean up endpoints. Also use when copying and deploying a 1P Tuned Model. Don't use for pure listing/discovery questions of the form "is X deployed?", "list my endpoints", or "which regions have models running?" — for those use `agent-platform-endpoint-management`. Don't use for public Vertex AI deployments (use the `vertex-deploy` skill) or for running model evaluations (use the `agent-platform-eval-flywheel` skill).

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

vaila-multimodaltoolbox/vaila212026년 10월 19일 업데이트

Manages Agent Platform serving endpoints. Use when you need to create, list, describe, update, or delete serving endpoints for model deployment on Agent Platform. Also use when troubleshooting endpoint permission, quota, or resource busy errors. Don't use for deploying models to endpoints or for running model evaluations.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

vaila-multimodaltoolbox/vaila212026년 10월 19일 업데이트

Measures and improves the quality of AI models and agents on Google Cloud using the Eval Quality Flywheel methodology. Use when evaluating an agent or model, building an eval dataset, picking or writing evaluation metrics, analyzing failures, comparing results before and after a fix, or when guidance is needed on Agent Platform eval methodology — including dataset schema, LLM-as-judge scoring, and common failure causes. For fine-tuning, use agent-platform-tuning. For general production deployment, use agent-platform-deploy.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

vaila-multimodaltoolbox/vaila212026년 10월 19일 업데이트

Connects to and performs inference with Google Cloud Agent Platform GenAI models, including First-Party Gemini models and Third-Party OpenMaaS models (Llama, DeepSeek, Qwen, etc.). Use when asked to perform inference, ask a model a question, run a test prompt, execute chat completions, or generate code for calling Gemini or OpenMaaS models, authenticate with GenAI SDK, OpenAI SDK, or legacy Agent Platform SDK, configure base URLs and global/regional endpoints, or troubleshoot 429 Resource Exhausted (DSQ), 400 User Validation, or 404 Not Found errors. Don't use for deploying models to endpoints or for running model evaluations.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

vaila-multimodaltoolbox/vaila212026년 10월 19일 업데이트

Guides agents and users through migrating from Gemini API in Google AI Studio to Gemini Enterprise Agent Platform (formerly Vertex AI). Use this skill when moving applications to Google Cloud, to leverage Cloud credits, or to unify inferencing with other Cloud infrastructure (IAM, billing, telemetry).

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

vaila-multimodaltoolbox/vaila212026년 10월 19일 업데이트

vaila-multimodaltoolbox의 스킬 모두 보기

이 스킬의 문제 신고