[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M69944Z3RAZ7MW3XRXWS9QC7":3,"skill-same:01M69944Z3RAZ7MW3XRXWS9QC7":76},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":39,"linkOnly":40,"markdownContent":41,"markdownBody":42,"frontMatter":43,"refName":47,"commitSha":48,"skillDirectory":49,"forksCount":50,"repositoryHtmlUrl":51,"repositoryHomepageUrl":6,"ownerHtmlUrl":52,"ownerBlogUrl":53,"ownerDescription":54,"updatedAt":55,"install":56,"resources":69},"01M69944Z3RAZ7MW3XRXWS9QC7","gke-inference",null,"Deploys and optimizes AI\u002FML inference workloads on GKE, using GPUs, TPUs, and model servers. Use when deploying GKE inference servers, configuring GKE GPU resources for inference, or deploying LLMs on GKE. Don't use for migrating existing AI workloads to GKE (use google-cloud-solution-guided-gke-ai-migration), GKE RAG with Cloud SQL\u002FAlloyDB (use google-cloud-solution-rag-enterprise-search-gke-sqldb), or batch\u002FHPC (use gke-batch-hpc).","google\u002Fskills","google","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F1342004?v=4","01M698VYEFDM0VJBZDYYJ2T83S","Google","01M698WR0H76ENHE587T9EJT89",24235,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgke-inference\u002FSKILL.md",[],[19,20,21,22,23,24,25,26,27,28,29,30,31,9,32,33,34,35,36,37,38],"inference","gke inference","gke","Enterprise","when","hpc","Existing","for","with","resources","use","model","google-cloud","batch","search","gpu","rag","ml","containers","llms","2026-11-10 10:12:30",false,"---\nname: gke-inference\ndescription: >-\n  Deploys and optimizes AI\u002FML inference workloads on GKE, using GPUs, TPUs, and\n  model servers. Use when deploying GKE inference servers, configuring GKE GPU\n  resources for inference, or deploying LLMs on GKE. Don't use for migrating\n  existing AI workloads to GKE (use google-cloud-solution-guided-gke-ai-migration),\n  GKE RAG with Cloud SQL\u002FAlloyDB (use google-cloud-solution-rag-enterprise-search-gke-sqldb),\n  or batch\u002FHPC (use gke-batch-hpc).\nmetadata:\n  version: \"1.0.2\"\n  category: Containers\n---\n\n# GKE AI\u002FML Inference\n\n> **Routing Note:** For migrating existing AI workloads to GKE, open `google-cloud-solution-guided-gke-ai-migration\u002FSKILL.md`. For GKE RAG with Cloud SQL or AlloyDB (`pgvector`), open `google-cloud-solution-rag-enterprise-search-gke-sqldb\u002FSKILL.md`.\n\nThis reference covers deploying AI\u002FML inference workloads on GKE using Google's\nInference Quickstart (GIQ) and best practices for LLM serving.\n\n> **MCP Tools:** `apply_k8s_manifest`, `get_k8s_resource`, `get_k8s_logs`,\n> `get_k8s_rollout_status`, `describe_k8s_resource`, `list_k8s_events`.\n> **CLI-only:** `gcloud container ai profiles *`\n\n## Attribution\n\n**MUST READ**: Before executing any `gcloud`, REST API, or MCP command, you\nMUST read [Usage Attribution](references\u002Fattribution.md). For commands executed\ndirectly in the terminal (never in generated scripts, Terraform, or user code),\nprefix `gcloud` with\n`CLOUDSDK_METRICS_REQUEST_ATTRIBUTION=\"agent-skills\u002F1.0 (skill:gke-inference)\"`\nand append `agent-skills\u002F1.0 (skill:gke-inference)` to `User-Agent` headers.\n\n## When to Use\n\n-   Deploy an AI model (Llama, Gemma, Mistral, etc.) to GKE\n-   Generate optimized Kubernetes manifests for inference\n-   Select GPU\u002FTPU accelerators for model serving\n-   Configure autoscaling for LLM inference\n\n## Prerequisites\n\n-   A golden path GKE Autopilot cluster (GPU workloads are supported via\n    ComputeClasses and NAP)\n-   `gcloud` CLI authenticated\n-   Sufficient GPU\u002FTPU quota in the target region\n\n## Workflow\n\n### 1. Discovery: Find Models and Hardware\n\n```bash\n# List all supported models\ngcloud container ai profiles models list --quiet\n\n# Find valid accelerator\u002Fserver combinations for a model\ngcloud container ai profiles list --model=\u003CMODEL_NAME> --quiet\n\n# Example: what can run Gemma 2 9B?\ngcloud container ai profiles list --model=gemma-2-9b-it --quiet\n```\n\n### 2. Generate Manifest\n\n```bash\ngcloud container ai profiles manifests create \\\n  --model=\u003CMODEL_NAME> \\\n  --model-server=\u003CSERVER> \\\n  --accelerator-type=\u003CACCELERATOR> \\\n  --target-ntpot-milliseconds=\u003CNTPOT> --quiet > inference.yaml\n```\n\n**Parameters:**\n\n-   `--model`: Model ID (e.g., `gemma-2-9b-it`, `llama-3-8b`)\n-   `--model-server`: Inference server (`vllm`, `tgi`, `triton`, `tensorrt-llm`)\n-   `--accelerator-type`: GPU\u002FTPU type (`nvidia-l4`, `nvidia-tesla-a100`,\n    `nvidia-h100-80gb`)\n-   `--target-ntpot-milliseconds`: Target Normalized Time Per Output Token\n    (optional, for latency optimization)\n\n**Example:**\n\n```bash\ngcloud container ai profiles manifests create \\\n  --model=gemma-2-9b-it \\\n  --model-server=vllm \\\n  --accelerator-type=nvidia-l4 \\\n  --target-ntpot-milliseconds=50 --quiet > inference.yaml\n```\n\n### 3. Review and Deploy\n\n```bash\n# Review for placeholders (HF tokens, PVCs)\ncat inference.yaml\n\n# Deploy\nkubectl apply -f inference.yaml\n\n# Monitor\nkubectl get pods -w\nkubectl logs -f \u003CPOD_NAME>\n```\n\n> Some models require Hugging Face tokens. Create a Kubernetes Secret and\n> reference it in the manifest.\n\n## GPU ComputeClass for Inference\n\nFor Autopilot clusters, create a ComputeClass to target GPU nodes:\n\n```yaml\napiVersion: cloud.google.com\u002Fv1\nkind: ComputeClass\nmetadata:\n  name: l4-inference\nspec:\n  priorities:\n  - machineFamily: g2\n    gpu:\n      type: nvidia-l4\n      count: 1\n    minCores: 4\n    minMemoryGb: 16\n```\n\n## Accelerator Selection Guide\n\n| Accelerator         | Best For                 | Memory      | Relative Cost |\n| ------------------- | ------------------------ | ----------- | ------------- |\n| NVIDIA T4           | Budget inference,        | 16 GB       | Lowest        |\n:                     : lightweight legacy       :             :               :\n:                     : models                   :             :               :\n| NVIDIA L4 (G2)      | Small-medium model       | 24 GB       | Low           |\n:                     : inference, video,        :             :               :\n:                     : graphics                 :             :               :\n| NVIDIA RTX PRO 6000 | Multimodal AI,           | 96 GB       | Medium        |\n: (G4)                : high-fidelity 3D,        :             :               :\n:                     : fine-tuning              :             :               :\n| Cloud TPU v5e       | Cost-effective           | Varies      | Medium        |\n:                     : transformer inference    :             :               :\n| Cloud TPU v5p       | High-performance         | Varies      | High          |\n:                     : training                 :             :               :\n| Cloud TPU v6e       | High-efficiency next-gen | 32 GB\u002Fchip  | Medium-High   |\n: (Trillium)          : training & serving       :             :               :\n| Cloud TPU v7x       | Ultra-scale inference &  | 192 GB\u002Fchip | High          |\n: (Ironwood)          : agentic workflows        :             :               :\n| NVIDIA A100         | Large model inference,   | 40\u002F80 GB    | High          |\n:                     : enterprise ML            :             :               :\n| NVIDIA H100 \u002F H200  | Frontier model training, | 80\u002F141 GB   | Highest       |\n:                     : high throughput          :             :               :\n| NVIDIA B200 (A4)    | Blackwell-scale          | 192 GB      | Highest       |\n:                     : training, FP4 precision  :             :               :\n| NVIDIA GB200 (A4X)  | Rack-scale AI (Grace     | Massive     | Highest       |\n:                     : Blackwell Superchip)     :             :               :\n\n## Autoscaling LLM Inference\n\n### GPU-based autoscaling\n\nUse custom metrics for GPU utilization:\n\n```yaml\napiVersion: autoscaling\u002Fv2\nkind: HorizontalPodAutoscaler\nmetadata:\n  name: llm-hpa\nspec:\n  scaleTargetRef:\n    apiVersion: apps\u002Fv1\n    kind: Deployment\n    name: llm-server\n  minReplicas: 1\n  maxReplicas: 10\n  metrics:\n  - type: Pods\n    pods:\n      metric:\n        name: gpu_duty_cycle\n      target:\n        type: AverageValue\n        averageValue: \"80\"\n```\n\n### Best practices for inference autoscaling\n\n1.  **Use DCGM metrics**: Golden path enables DCGM monitoring for GPU\n    utilization metrics\n2.  **Set appropriate minReplicas**: At least 1 for always-on serving; 0 for\n    batch\u002Fon-demand\n3.  **Tune scale-down delay**: LLM model loading is slow; use longer\n    stabilization windows\n4.  **Consider queue depth**: Scale on pending requests rather than pure GPU\n    utilization for latency-sensitive workloads\n\n## Optimization Tips\n\n-   **Quantization**: Use quantized models (GPTQ, AWQ) to reduce GPU memory and\n    increase throughput\n-   **Batching**: Configure model server batch size for throughput vs latency\n    trade-off\n-   **Tensor parallelism**: Split large models across multiple GPUs within a\n    node\n-   **KV cache optimization**: Tune `--gpu-memory-utilization` in vLLM for KV\n    cache allocation\n\n## Troubleshooting\n\n| Issue              | Cause                    | Fix                         |\n| ------------------ | ------------------------ | --------------------------- |\n| Invalid            | Unsupported tuple        | Re-run `gcloud container ai |\n: model\u002Faccelerator  :                          : profiles list               :\n: combination        :                          : --model=\u003CMODEL>`            :\n| GPU quota exceeded | Regional quota limit     | Request quota increase or   |\n:                    :                          : try a different region      :\n| OOM on GPU         | Model too large for      | Use larger GPU, enable      |\n:                    : accelerator              : quantization, or use tensor :\n:                    :                          : parallelism                 :\n| Slow cold start    | Large model loading from | Use local SSD for model     |\n:                    : registry                 : caching; pre-pull images    :\n","# GKE AI\u002FML Inference\n\n> **Routing Note:** For migrating existing AI workloads to GKE, open `google-cloud-solution-guided-gke-ai-migration\u002FSKILL.md`. For GKE RAG with Cloud SQL or AlloyDB (`pgvector`), open `google-cloud-solution-rag-enterprise-search-gke-sqldb\u002FSKILL.md`.\n\nThis reference covers deploying AI\u002FML inference workloads on GKE using Google's\nInference Quickstart (GIQ) and best practices for LLM serving.\n\n> **MCP Tools:** `apply_k8s_manifest`, `get_k8s_resource`, `get_k8s_logs`,\n> `get_k8s_rollout_status`, `describe_k8s_resource`, `list_k8s_events`.\n> **CLI-only:** `gcloud container ai profiles *`\n\n## Attribution\n\n**MUST READ**: Before executing any `gcloud`, REST API, or MCP command, you\nMUST read [Usage Attribution](references\u002Fattribution.md). For commands executed\ndirectly in the terminal (never in generated scripts, Terraform, or user code),\nprefix `gcloud` with\n`CLOUDSDK_METRICS_REQUEST_ATTRIBUTION=\"agent-skills\u002F1.0 (skill:gke-inference)\"`\nand append `agent-skills\u002F1.0 (skill:gke-inference)` to `User-Agent` headers.\n\n## When to Use\n\n-   Deploy an AI model (Llama, Gemma, Mistral, etc.) to GKE\n-   Generate optimized Kubernetes manifests for inference\n-   Select GPU\u002FTPU accelerators for model serving\n-   Configure autoscaling for LLM inference\n\n## Prerequisites\n\n-   A golden path GKE Autopilot cluster (GPU workloads are supported via\n    ComputeClasses and NAP)\n-   `gcloud` CLI authenticated\n-   Sufficient GPU\u002FTPU quota in the target region\n\n## Workflow\n\n### 1. Discovery: Find Models and Hardware\n\n```bash\n# List all supported models\ngcloud container ai profiles models list --quiet\n\n# Find valid accelerator\u002Fserver combinations for a model\ngcloud container ai profiles list --model=\u003CMODEL_NAME> --quiet\n\n# Example: what can run Gemma 2 9B?\ngcloud container ai profiles list --model=gemma-2-9b-it --quiet\n```\n\n### 2. Generate Manifest\n\n```bash\ngcloud container ai profiles manifests create \\\n  --model=\u003CMODEL_NAME> \\\n  --model-server=\u003CSERVER> \\\n  --accelerator-type=\u003CACCELERATOR> \\\n  --target-ntpot-milliseconds=\u003CNTPOT> --quiet > inference.yaml\n```\n\n**Parameters:**\n\n-   `--model`: Model ID (e.g., `gemma-2-9b-it`, `llama-3-8b`)\n-   `--model-server`: Inference server (`vllm`, `tgi`, `triton`, `tensorrt-llm`)\n-   `--accelerator-type`: GPU\u002FTPU type (`nvidia-l4`, `nvidia-tesla-a100`,\n    `nvidia-h100-80gb`)\n-   `--target-ntpot-milliseconds`: Target Normalized Time Per Output Token\n    (optional, for latency optimization)\n\n**Example:**\n\n```bash\ngcloud container ai profiles manifests create \\\n  --model=gemma-2-9b-it \\\n  --model-server=vllm \\\n  --accelerator-type=nvidia-l4 \\\n  --target-ntpot-milliseconds=50 --quiet > inference.yaml\n```\n\n### 3. Review and Deploy\n\n```bash\n# Review for placeholders (HF tokens, PVCs)\ncat inference.yaml\n\n# Deploy\nkubectl apply -f inference.yaml\n\n# Monitor\nkubectl get pods -w\nkubectl logs -f \u003CPOD_NAME>\n```\n\n> Some models require Hugging Face tokens. Create a Kubernetes Secret and\n> reference it in the manifest.\n\n## GPU ComputeClass for Inference\n\nFor Autopilot clusters, create a ComputeClass to target GPU nodes:\n\n```yaml\napiVersion: cloud.google.com\u002Fv1\nkind: ComputeClass\nmetadata:\n  name: l4-inference\nspec:\n  priorities:\n  - machineFamily: g2\n    gpu:\n      type: nvidia-l4\n      count: 1\n    minCores: 4\n    minMemoryGb: 16\n```\n\n## Accelerator Selection Guide\n\n| Accelerator         | Best For                 | Memory      | Relative Cost |\n| ------------------- | ------------------------ | ----------- | ------------- |\n| NVIDIA T4           | Budget inference,        | 16 GB       | Lowest        |\n:                     : lightweight legacy       :             :               :\n:                     : models                   :             :               :\n| NVIDIA L4 (G2)      | Small-medium model       | 24 GB       | Low           |\n:                     : inference, video,        :             :               :\n:                     : graphics                 :             :               :\n| NVIDIA RTX PRO 6000 | Multimodal AI,           | 96 GB       | Medium        |\n: (G4)                : high-fidelity 3D,        :             :               :\n:                     : fine-tuning              :             :               :\n| Cloud TPU v5e       | Cost-effective           | Varies      | Medium        |\n:                     : transformer inference    :             :               :\n| Cloud TPU v5p       | High-performance         | Varies      | High          |\n:                     : training                 :             :               :\n| Cloud TPU v6e       | High-efficiency next-gen | 32 GB\u002Fchip  | Medium-High   |\n: (Trillium)          : training & serving       :             :               :\n| Cloud TPU v7x       | Ultra-scale inference &  | 192 GB\u002Fchip | High          |\n: (Ironwood)          : agentic workflows        :             :               :\n| NVIDIA A100         | Large model inference,   | 40\u002F80 GB    | High          |\n:                     : enterprise ML            :             :               :\n| NVIDIA H100 \u002F H200  | Frontier model training, | 80\u002F141 GB   | Highest       |\n:                     : high throughput          :             :               :\n| NVIDIA B200 (A4)    | Blackwell-scale          | 192 GB      | Highest       |\n:                     : training, FP4 precision  :             :               :\n| NVIDIA GB200 (A4X)  | Rack-scale AI (Grace     | Massive     | Highest       |\n:                     : Blackwell Superchip)     :             :               :\n\n## Autoscaling LLM Inference\n\n### GPU-based autoscaling\n\nUse custom metrics for GPU utilization:\n\n```yaml\napiVersion: autoscaling\u002Fv2\nkind: HorizontalPodAutoscaler\nmetadata:\n  name: llm-hpa\nspec:\n  scaleTargetRef:\n    apiVersion: apps\u002Fv1\n    kind: Deployment\n    name: llm-server\n  minReplicas: 1\n  maxReplicas: 10\n  metrics:\n  - type: Pods\n    pods:\n      metric:\n        name: gpu_duty_cycle\n      target:\n        type: AverageValue\n        averageValue: \"80\"\n```\n\n### Best practices for inference autoscaling\n\n1.  **Use DCGM metrics**: Golden path enables DCGM monitoring for GPU\n    utilization metrics\n2.  **Set appropriate minReplicas**: At least 1 for always-on serving; 0 for\n    batch\u002Fon-demand\n3.  **Tune scale-down delay**: LLM model loading is slow; use longer\n    stabilization windows\n4.  **Consider queue depth**: Scale on pending requests rather than pure GPU\n    utilization for latency-sensitive workloads\n\n## Optimization Tips\n\n-   **Quantization**: Use quantized models (GPTQ, AWQ) to reduce GPU memory and\n    increase throughput\n-   **Batching**: Configure model server batch size for throughput vs latency\n    trade-off\n-   **Tensor parallelism**: Split large models across multiple GPUs within a\n    node\n-   **KV cache optimization**: Tune `--gpu-memory-utilization` in vLLM for KV\n    cache allocation\n\n## Troubleshooting\n\n| Issue              | Cause                    | Fix                         |\n| ------------------ | ------------------------ | --------------------------- |\n| Invalid            | Unsupported tuple        | Re-run `gcloud container ai |\n: model\u002Faccelerator  :                          : profiles list               :\n: combination        :                          : --model=\u003CMODEL>`            :\n| GPU quota exceeded | Regional quota limit     | Request quota increase or   |\n:                    :                          : try a different region      :\n| OOM on GPU         | Model too large for      | Use larger GPU, enable      |\n:                    : accelerator              : quantization, or use tensor :\n:                    :                          : parallelism                 :\n| Slow cold start    | Large model loading from | Use local SSD for model     |\n:                    : registry                 : caching; pre-pull images    :\n",{"name":5,"metadata":44,"description":7},{"version":45,"category":46},"1.0.2","Containers","main","bb9b3f464aae37af360d079c1d863524955b0363","skills\u002Fcloud\u002Fgke-inference",2014,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills","https:\u002F\u002Fgithub.com\u002Fgoogle","https:\u002F\u002Fopensource.google\u002F","Google ❤️ Open Source","2026-11-14 23:31:00",{"source":8,"skillName":5,"commands":57,"prompt":67,"zipAvailable":68},[58,61,64],{"tool":59,"command":60},"npx","npx skills add google\u002Fskills --skill gke-inference",{"tool":62,"command":63},"bunx","bunx skills add google\u002Fskills --skill gke-inference",{"tool":65,"command":66},"pnpm","pnpm dlx skills add google\u002Fskills --skill gke-inference","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgke-inference\u002FSKILL.md",true,[70],{"path":71,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":74,"sourceHtmlUrl":75},"references\u002Fattribution.md","REFERENCE","参考文档",1879,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgke-inference\u002Freferences\u002Fattribution.md",[77,85,92,99,106,113],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":84},"01M7DQ9K7JF4T7NJZ2K1MQZ44Q","cloud-operator-share-context","Shares Google Cloud operating context the user or their team already have — environments, projects, services and regions, product preferences, best practices, architecture guidance, runbooks — found in agent instruction files, local memory, skills and workflows, and the repository's GCP docs, with the operator agent's memory after the user reviews the list. Use when the user runs this skill, accepts the cloud-operator or cloud-operator-setup skill's offer to share their existing Google Cloud context, or asks to share, import or save that context or a GCP fact for the operator agent. Don't use for ordinary GCP requests, which go to the cloud-operator skill.","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fcloud-operator-share-context\u002FSKILL.md",[],[],"2026-11-15 14:19:17",{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":84},"01M7DQ9ERFGGQCFAWQW0KRKYBY","cloud-operator-setup","Sets up the operator agent in a Google Cloud project via the autocloud MCP server: picks the operator project, creates or checks the user's operator agent, saves the project, and offers to share the user's existing Google Cloud operating context with the operator agent's memory. Use when the user runs this skill, asks to set up the operator or create their operator agent, or chooses setup when the cloud-operator skill asks. Don't use for ordinary GCP requests or installing gcloud.","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fcloud-operator-setup\u002FSKILL.md",[],[],{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":96,"categories":97,"keywords":98,"publishedAt":84},"01M7B4RHQCM0DK2XQ7S7ZXVN8D","developer-device-platform-device-run","Runs automated, non-interactive jobs on remote Android devices with Developer Device Platform (DDP) Device Run using `gcloud beta device-run sessions`: native Android executables or shell scripts (for example CLI tools, custom runners, test harnesses or benchmarks) and Android instrumentation tests, on one or many devices, then waits for completion and collects per-device outputs and logs. Use when running a binary or test suite on remote devices without an interactive connection. Don't use it for interactive device reservations, adb tunnels or screen streaming (use developer-device-platform-basics), or for local devices.","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fdeveloper-device-platform-device-run\u002FSKILL.md",[],[],{"skillNo":100,"name":101,"nameZh":6,"description":102,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":84},"01M7B4R9YXMRW7RZP9Q9FMZJ2P","compute-vm-creation","Guides the planning, configuration, and provisioning of standalone Google Compute Engine virtual machines (VMs) and bare metal (-metal) instances using gcloud CLI commands or direct MCP tool execution. Use ONLY when the user wants to create, configure, or provision a new standalone Compute Engine VM or bare metal instance from conversational intents (hardware sizing, region selection, OS image, networking, security defaults). DO NOT read or use for Google Kubernetes Engine (GKE) Autopilot or Standard clusters, Kubernetes node pools or Pods, container orchestration, managing existing VMs, Managed Instance Groups (MIGs), or Instance Templates.","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fcompute-vm-creation\u002FSKILL.md",[],[],{"skillNo":107,"name":108,"nameZh":6,"description":109,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":110,"categories":111,"keywords":112,"publishedAt":84},"01M7B4QD02N0YBVT0838Z5K5VC","agent-platform-autorater-customization","Customizes an autorater (evaluation metric) for an AI agent so its grades match the user's own judgment, starting from the agent's real trajectories. Use when a user wants a metric tailored to their agent, wants to turn failing trajectories into rubrics, says an existing autorater grades their agent wrong, or wants an autorater to agree with human labels (autorater hillclimbing, judge calibration). Don't use for quick evals with built-in metrics or for improving the agent itself (use agent-platform-eval-flywheel).","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fagent-platform-autorater-customization\u002FSKILL.md",[],[],{"skillNo":114,"name":115,"nameZh":6,"description":116,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":117,"categories":118,"keywords":119,"publishedAt":84},"01M78JP22K73KK96JH0JNDT5C4","iap-agent-platform-troubleshooting","Diagnoses Identity-Aware Proxy (IAP) denials of outbound requests from Gemini Enterprise Agent Platform agents that run on Agent Runtime and send traffic through Agent Gateway. Use when an agent request fails with \"Egress request is not authorized\", when Cloud Audit Logs show an IAP egressViaIAP decision with granted false, when IAP audit entries are missing while debugging agent egress, or when Access policies, IAP allow policies, policy bindings, or Agent Registry entries might block agent egress. Don't use for Gemini model IAM permission errors (such as aiplatform.endpoints.predict), Agent Platform API errors, or Agent Runtime deployment or 500 errors (use agent-platform-troubleshooting instead). Don't use for setting up, creating, or connecting an Agent Gateway, or for IAP web apps, IAP TCP forwarding, or SSH.","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fiap-agent-platform-troubleshooting\u002FSKILL.md",[],[]]