[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6P86GSH66C1K45Z3D3V70H1":3,"skill-same:01M6P86GSH66C1K45Z3D3V70H1":71},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":38,"linkOnly":39,"markdownContent":40,"markdownBody":41,"frontMatter":42,"refName":22,"commitSha":52,"skillDirectory":53,"forksCount":13,"repositoryHtmlUrl":54,"repositoryHomepageUrl":6,"ownerHtmlUrl":55,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":56,"install":57,"resources":70},"01M6P86GSH66C1K45Z3D3V70H1","llm-inference-scaling",null,"Auto-scale LLM inference clusters on Kubernetes using KEDA, custom GPU metrics, and horizontal pod autoscaling.","duc-duy-1612\u002FOnline-Delivery-Systems-Web","duc-duy-1612","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F192231384?v=4","01M6NSDKQQDKVTHZ8DZMZ1VYY6","01M6NSDKW8DDVTQHBT1HBDDJPT",0,"MIT","https:\u002F\u002Fgithub.com\u002Fduc-duy-1612\u002FOnline-Delivery-Systems-Web\u002Fblob\u002Fmain\u002F.agents\u002Fskills_archive\u002Fllm-inference-scaling\u002FSKILL.md",[],[18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37],"inference","llm inference scaling","llm inference","scaling","main","autoscaling","keda","relevant","scripts","critical","os","platform","kubernetes","community","only","auto","access","devops-security","metrics","security","2026-11-10 10:12:30",false,"---\nname: llm-inference-scaling\ndescription: Auto-scale LLM inference clusters on Kubernetes using KEDA, custom GPU\n  metrics, and horizontal pod autoscaling.\ncategory: devops\nrisk: critical\nsource: https:\u002F\u002Fgithub.com\u002FBagelHole\u002FDevOps-Security-Agent-Skills\nsource_repo: BagelHole\u002FDevOps-Security-Agent-Skills\nsource_type: community\ndate_added: '2026-09-20'\nlicense: MIT\nlicense_source: https:\u002F\u002Fgithub.com\u002FBagelHole\u002FDevOps-Security-Agent-Skills\u002Fblob\u002Fmain\u002FLICENSE\ncompatibility: Requires the relevant OS\u002Fplatform tooling and privileged access where\n  noted. Docs-only; helper scripts and templates not bundled.\nmetadata:\n  author: devops-skills\n  version: '1.0'\n---\n\n# LLM Inference Scaling\n\nScale LLM inference horizontally on Kubernetes with GPU-aware autoscaling, request queuing, and cost-efficient spot instance strategies.\n\n## When to Use This Skill\n\nUse this skill when:\n- LLM API traffic is unpredictable and you need to scale up\u002Fdown automatically\n- Managing a fleet of vLLM or TGI inference pods on Kubernetes\n- Reducing inference costs with spot\u002Fpreemptible GPU instances\n- Implementing queue-based autoscaling for batch inference jobs\n- Building a multi-model serving platform that shares GPU resources\n\n## Prerequisites\n\n- Kubernetes cluster with GPU nodes (NVIDIA operator installed)\n- KEDA (Kubernetes Event-Driven Autoscaler) installed\n- Prometheus with GPU metrics (`dcgm-exporter` or `gpu-operator`)\n- Helm 3+ for chart deployments\n\n## GPU Node Setup\n\n```bash\n# Install NVIDIA GPU Operator (handles drivers, container toolkit, DCGM)\nhelm repo add nvidia https:\u002F\u002Fhelm.ngc.nvidia.com\u002Fnvidia\nhelm repo update\n\nhelm install gpu-operator nvidia\u002Fgpu-operator \\\n  --namespace gpu-operator \\\n  --create-namespace \\\n  --set driver.enabled=true \\\n  --set dcgm.enabled=true \\\n  --set devicePlugin.enabled=true\n\n# Verify GPU nodes are recognized\nkubectl get nodes -l nvidia.com\u002Fgpu.present=true\nkubectl describe node \u003Cgpu-node> | grep nvidia\n```\n\n## vLLM Deployment with GPU Resources\n\n```yaml\napiVersion: apps\u002Fv1\nkind: Deployment\nmetadata:\n  name: vllm-llama-8b\n  labels:\n    app: vllm\n    model: llama-3.1-8b\nspec:\n  replicas: 1\n  selector:\n    matchLabels:\n      app: vllm\n      model: llama-3.1-8b\n  template:\n    metadata:\n      labels:\n        app: vllm\n        model: llama-3.1-8b\n    spec:\n      nodeSelector:\n        nvidia.com\u002Fgpu.present: \"true\"\n      tolerations:\n      - key: nvidia.com\u002Fgpu\n        operator: Exists\n        effect: NoSchedule\n      containers:\n      - name: vllm\n        image: vllm\u002Fvllm-openai:latest\n        args:\n        - \"--model\"\n        - \"meta-llama\u002FLlama-3.1-8B-Instruct\"\n        - \"--tensor-parallel-size\"\n        - \"1\"\n        - \"--gpu-memory-utilization\"\n        - \"0.90\"\n        - \"--max-num-seqs\"\n        - \"128\"\n        resources:\n          requests:\n            nvidia.com\u002Fgpu: \"1\"\n            memory: \"20Gi\"\n            cpu: \"4\"\n          limits:\n            nvidia.com\u002Fgpu: \"1\"\n            memory: \"24Gi\"\n            cpu: \"8\"\n        ports:\n        - containerPort: 8000\n        readinessProbe:\n          httpGet:\n            path: \u002Fhealth\n            port: 8000\n          initialDelaySeconds: 60\n          periodSeconds: 10\n        env:\n        - name: HUGGING_FACE_HUB_TOKEN\n          valueFrom:\n            secretKeyRef:\n              name: hf-token\n              key: token\n```\n\n## KEDA Autoscaling on Prometheus Metrics\n\n```yaml\napiVersion: keda.sh\u002Fv1alpha1\nkind: ScaledObject\nmetadata:\n  name: vllm-scaledobject\nspec:\n  scaleTargetRef:\n    name: vllm-llama-8b\n  minReplicaCount: 1\n  maxReplicaCount: 8\n  cooldownPeriod: 300          # 5 min before scale-down\n  pollingInterval: 15\n  triggers:\n  - type: prometheus\n    metadata:\n      serverAddress: http:\u002F\u002Fprometheus-server.monitoring:9090\n      metricName: vllm_num_requests_waiting\n      threshold: \"10\"           # scale up if >10 requests waiting\n      query: |\n        sum(vllm:num_requests_waiting{deployment=\"vllm-llama-8b\"})\n  - type: prometheus\n    metadata:\n      serverAddress: http:\u002F\u002Fprometheus-server.monitoring:9090\n      metricName: vllm_gpu_cache_usage\n      threshold: \"0.8\"          # scale up if KV cache >80% full\n      query: |\n        avg(vllm:gpu_cache_usage_perc{deployment=\"vllm-llama-8b\"})\n```\n\n## Queue-Based Scaling (Redis + KEDA)\n\n```yaml\n# ScaledJob for async batch inference\napiVersion: keda.sh\u002Fv1alpha1\nkind: ScaledJob\nmetadata:\n  name: llm-batch-inference\nspec:\n  jobTargetRef:\n    template:\n      spec:\n        containers:\n        - name: inference-worker\n          image: myapp\u002Finference-worker:latest\n          env:\n          - name: REDIS_URL\n            value: redis:\u002F\u002Fredis:6379\n          - name: QUEUE_NAME\n            value: inference-jobs\n        restartPolicy: OnFailure\n  minReplicaCount: 0\n  maxReplicaCount: 20\n  pollingInterval: 5\n  successfulJobsHistoryLimit: 3\n  triggers:\n  - type: redis\n    metadata:\n      address: redis:6379\n      listName: inference-jobs\n      listLength: \"5\"           # 1 worker per 5 queued jobs\n```\n\n## Spot Instance Strategy\n\n```yaml\n# Mixed node pool: on-demand + spot GPUs\napiVersion: v1\nkind: ConfigMap\nmetadata:\n  name: cluster-autoscaler-priority-config\ndata:\n  priorities: |\n    10:  # low priority = prefer\n    - .*spot.*\n    50:\n    - .*on-demand.*\n---\n# Node affinity for spot with on-demand fallback\nspec:\n  affinity:\n    nodeAffinity:\n      preferredDuringSchedulingIgnoredDuringExecution:\n      - weight: 80\n        preference:\n          matchExpressions:\n          - key: node.kubernetes.io\u002Flifecycle\n            operator: In\n            values: [spot]\n      - weight: 20\n        preference:\n          matchExpressions:\n          - key: node.kubernetes.io\u002Flifecycle\n            operator: In\n            values: [on-demand]\n```\n\n## Cluster Autoscaler for GPU Nodes\n\n```bash\n# AWS EKS — enable cluster autoscaler for GPU node group\nhelm install cluster-autoscaler autoscaler\u002Fcluster-autoscaler \\\n  --namespace kube-system \\\n  --set autoDiscovery.clusterName=my-cluster \\\n  --set awsRegion=us-east-1 \\\n  --set rbac.serviceAccount.annotations.\"eks\\.amazonaws\\.com\u002Frole-arn\"=arn:aws:iam::ACCOUNT:role\u002FClusterAutoscalerRole \\\n  --set extraArgs.skip-nodes-with-local-storage=false \\\n  --set extraArgs.expander=least-waste\n\n# Annotate GPU node group for autoscaler\nkubectl annotate node \u003Cnode> \\\n  cluster-autoscaler.kubernetes.io\u002Fsafe-to-evict=\"false\"\n```\n\n## Scaling Metrics to Monitor\n\n```bash\n# Prometheus queries for scaling decisions\n# Requests waiting in vLLM queue\nsum(vllm:num_requests_waiting) by (model)\n\n# GPU KV cache utilization (>80% = bottleneck)\navg(vllm:gpu_cache_usage_perc) by (pod)\n\n# Tokens per second throughput\nsum(rate(vllm:generation_tokens_total[5m])) by (model)\n\n# P99 time-to-first-token\nhistogram_quantile(0.99, rate(vllm:time_to_first_token_seconds_bucket[5m]))\n```\n\n## Common Issues\n\n| Issue | Cause | Fix |\n|-------|-------|-----|\n| Pods stuck in `Pending` | No GPU nodes available | Check cluster autoscaler logs; verify node group limits |\n| Scale-up too slow | Cluster autoscaler delay + model load time | Pre-warm replicas; increase `minReplicaCount` |\n| GPU fragmentation | Multiple small models on large GPUs | Use MIG partitioning or consolidate model sizes |\n| Spot eviction causes errors | Spot instance reclamation | Add `PodDisruptionBudget`; use graceful shutdown |\n| KEDA not scaling | Prometheus query returns no data | Test query in Prometheus UI first |\n\n## Best Practices\n\n- Set `minReplicaCount: 1` to avoid cold starts; scale to 0 only for batch jobs.\n- Use `PodDisruptionBudget` with `minAvailable: 1` to survive spot evictions.\n- Pre-pull model weights into a shared PVC to speed up pod startup by 5–10×.\n- Separate model families across node pools (A10G for 7B, A100 for 70B).\n- Use Kubernetes VPA for CPU\u002Fmemory right-sizing alongside KEDA for replica count.\n\n## Related Skills\n\n- vllm-server (`vllm-server`) - vLLM configuration and tuning\n- gpu-server-management (`gpu-server-management`) - GPU node setup\n- model-serving-kubernetes (`model-serving-kubernetes`) - KServe\n- kubernetes-ops (`kubernetes-ops`) - Core Kubernetes\n- llm-cost-optimization (`llm-cost-optimization`) - Cost strategies\n\n## Limitations\n\n- Infrastructure commands can disrupt services: confirm target host\u002Fscope and have backups\u002Fsnapshots before mutating state.\n- Docs-only import: upstream scripts and templates not bundled.\n\n","# LLM Inference Scaling\n\nScale LLM inference horizontally on Kubernetes with GPU-aware autoscaling, request queuing, and cost-efficient spot instance strategies.\n\n## When to Use This Skill\n\nUse this skill when:\n- LLM API traffic is unpredictable and you need to scale up\u002Fdown automatically\n- Managing a fleet of vLLM or TGI inference pods on Kubernetes\n- Reducing inference costs with spot\u002Fpreemptible GPU instances\n- Implementing queue-based autoscaling for batch inference jobs\n- Building a multi-model serving platform that shares GPU resources\n\n## Prerequisites\n\n- Kubernetes cluster with GPU nodes (NVIDIA operator installed)\n- KEDA (Kubernetes Event-Driven Autoscaler) installed\n- Prometheus with GPU metrics (`dcgm-exporter` or `gpu-operator`)\n- Helm 3+ for chart deployments\n\n## GPU Node Setup\n\n```bash\n# Install NVIDIA GPU Operator (handles drivers, container toolkit, DCGM)\nhelm repo add nvidia https:\u002F\u002Fhelm.ngc.nvidia.com\u002Fnvidia\nhelm repo update\n\nhelm install gpu-operator nvidia\u002Fgpu-operator \\\n  --namespace gpu-operator \\\n  --create-namespace \\\n  --set driver.enabled=true \\\n  --set dcgm.enabled=true \\\n  --set devicePlugin.enabled=true\n\n# Verify GPU nodes are recognized\nkubectl get nodes -l nvidia.com\u002Fgpu.present=true\nkubectl describe node \u003Cgpu-node> | grep nvidia\n```\n\n## vLLM Deployment with GPU Resources\n\n```yaml\napiVersion: apps\u002Fv1\nkind: Deployment\nmetadata:\n  name: vllm-llama-8b\n  labels:\n    app: vllm\n    model: llama-3.1-8b\nspec:\n  replicas: 1\n  selector:\n    matchLabels:\n      app: vllm\n      model: llama-3.1-8b\n  template:\n    metadata:\n      labels:\n        app: vllm\n        model: llama-3.1-8b\n    spec:\n      nodeSelector:\n        nvidia.com\u002Fgpu.present: \"true\"\n      tolerations:\n      - key: nvidia.com\u002Fgpu\n        operator: Exists\n        effect: NoSchedule\n      containers:\n      - name: vllm\n        image: vllm\u002Fvllm-openai:latest\n        args:\n        - \"--model\"\n        - \"meta-llama\u002FLlama-3.1-8B-Instruct\"\n        - \"--tensor-parallel-size\"\n        - \"1\"\n        - \"--gpu-memory-utilization\"\n        - \"0.90\"\n        - \"--max-num-seqs\"\n        - \"128\"\n        resources:\n          requests:\n            nvidia.com\u002Fgpu: \"1\"\n            memory: \"20Gi\"\n            cpu: \"4\"\n          limits:\n            nvidia.com\u002Fgpu: \"1\"\n            memory: \"24Gi\"\n            cpu: \"8\"\n        ports:\n        - containerPort: 8000\n        readinessProbe:\n          httpGet:\n            path: \u002Fhealth\n            port: 8000\n          initialDelaySeconds: 60\n          periodSeconds: 10\n        env:\n        - name: HUGGING_FACE_HUB_TOKEN\n          valueFrom:\n            secretKeyRef:\n              name: hf-token\n              key: token\n```\n\n## KEDA Autoscaling on Prometheus Metrics\n\n```yaml\napiVersion: keda.sh\u002Fv1alpha1\nkind: ScaledObject\nmetadata:\n  name: vllm-scaledobject\nspec:\n  scaleTargetRef:\n    name: vllm-llama-8b\n  minReplicaCount: 1\n  maxReplicaCount: 8\n  cooldownPeriod: 300          # 5 min before scale-down\n  pollingInterval: 15\n  triggers:\n  - type: prometheus\n    metadata:\n      serverAddress: http:\u002F\u002Fprometheus-server.monitoring:9090\n      metricName: vllm_num_requests_waiting\n      threshold: \"10\"           # scale up if >10 requests waiting\n      query: |\n        sum(vllm:num_requests_waiting{deployment=\"vllm-llama-8b\"})\n  - type: prometheus\n    metadata:\n      serverAddress: http:\u002F\u002Fprometheus-server.monitoring:9090\n      metricName: vllm_gpu_cache_usage\n      threshold: \"0.8\"          # scale up if KV cache >80% full\n      query: |\n        avg(vllm:gpu_cache_usage_perc{deployment=\"vllm-llama-8b\"})\n```\n\n## Queue-Based Scaling (Redis + KEDA)\n\n```yaml\n# ScaledJob for async batch inference\napiVersion: keda.sh\u002Fv1alpha1\nkind: ScaledJob\nmetadata:\n  name: llm-batch-inference\nspec:\n  jobTargetRef:\n    template:\n      spec:\n        containers:\n        - name: inference-worker\n          image: myapp\u002Finference-worker:latest\n          env:\n          - name: REDIS_URL\n            value: redis:\u002F\u002Fredis:6379\n          - name: QUEUE_NAME\n            value: inference-jobs\n        restartPolicy: OnFailure\n  minReplicaCount: 0\n  maxReplicaCount: 20\n  pollingInterval: 5\n  successfulJobsHistoryLimit: 3\n  triggers:\n  - type: redis\n    metadata:\n      address: redis:6379\n      listName: inference-jobs\n      listLength: \"5\"           # 1 worker per 5 queued jobs\n```\n\n## Spot Instance Strategy\n\n```yaml\n# Mixed node pool: on-demand + spot GPUs\napiVersion: v1\nkind: ConfigMap\nmetadata:\n  name: cluster-autoscaler-priority-config\ndata:\n  priorities: |\n    10:  # low priority = prefer\n    - .*spot.*\n    50:\n    - .*on-demand.*\n---\n# Node affinity for spot with on-demand fallback\nspec:\n  affinity:\n    nodeAffinity:\n      preferredDuringSchedulingIgnoredDuringExecution:\n      - weight: 80\n        preference:\n          matchExpressions:\n          - key: node.kubernetes.io\u002Flifecycle\n            operator: In\n            values: [spot]\n      - weight: 20\n        preference:\n          matchExpressions:\n          - key: node.kubernetes.io\u002Flifecycle\n            operator: In\n            values: [on-demand]\n```\n\n## Cluster Autoscaler for GPU Nodes\n\n```bash\n# AWS EKS — enable cluster autoscaler for GPU node group\nhelm install cluster-autoscaler autoscaler\u002Fcluster-autoscaler \\\n  --namespace kube-system \\\n  --set autoDiscovery.clusterName=my-cluster \\\n  --set awsRegion=us-east-1 \\\n  --set rbac.serviceAccount.annotations.\"eks\\.amazonaws\\.com\u002Frole-arn\"=arn:aws:iam::ACCOUNT:role\u002FClusterAutoscalerRole \\\n  --set extraArgs.skip-nodes-with-local-storage=false \\\n  --set extraArgs.expander=least-waste\n\n# Annotate GPU node group for autoscaler\nkubectl annotate node \u003Cnode> \\\n  cluster-autoscaler.kubernetes.io\u002Fsafe-to-evict=\"false\"\n```\n\n## Scaling Metrics to Monitor\n\n```bash\n# Prometheus queries for scaling decisions\n# Requests waiting in vLLM queue\nsum(vllm:num_requests_waiting) by (model)\n\n# GPU KV cache utilization (>80% = bottleneck)\navg(vllm:gpu_cache_usage_perc) by (pod)\n\n# Tokens per second throughput\nsum(rate(vllm:generation_tokens_total[5m])) by (model)\n\n# P99 time-to-first-token\nhistogram_quantile(0.99, rate(vllm:time_to_first_token_seconds_bucket[5m]))\n```\n\n## Common Issues\n\n| Issue | Cause | Fix |\n|-------|-------|-----|\n| Pods stuck in `Pending` | No GPU nodes available | Check cluster autoscaler logs; verify node group limits |\n| Scale-up too slow | Cluster autoscaler delay + model load time | Pre-warm replicas; increase `minReplicaCount` |\n| GPU fragmentation | Multiple small models on large GPUs | Use MIG partitioning or consolidate model sizes |\n| Spot eviction causes errors | Spot instance reclamation | Add `PodDisruptionBudget`; use graceful shutdown |\n| KEDA not scaling | Prometheus query returns no data | Test query in Prometheus UI first |\n\n## Best Practices\n\n- Set `minReplicaCount: 1` to avoid cold starts; scale to 0 only for batch jobs.\n- Use `PodDisruptionBudget` with `minAvailable: 1` to survive spot evictions.\n- Pre-pull model weights into a shared PVC to speed up pod startup by 5–10×.\n- Separate model families across node pools (A10G for 7B, A100 for 70B).\n- Use Kubernetes VPA for CPU\u002Fmemory right-sizing alongside KEDA for replica count.\n\n## Related Skills\n\n- vllm-server (`vllm-server`) - vLLM configuration and tuning\n- gpu-server-management (`gpu-server-management`) - GPU node setup\n- model-serving-kubernetes (`model-serving-kubernetes`) - KServe\n- kubernetes-ops (`kubernetes-ops`) - Core Kubernetes\n- llm-cost-optimization (`llm-cost-optimization`) - Cost strategies\n\n## Limitations\n\n- Infrastructure commands can disrupt services: confirm target host\u002Fscope and have backups\u002Fsnapshots before mutating state.\n- Docs-only import: upstream scripts and templates not bundled.\n\n",{"name":5,"risk":27,"source":43,"license":14,"category":44,"metadata":45,"date_added":48,"description":7,"source_repo":49,"source_type":31,"compatibility":50,"license_source":51},"https:\u002F\u002Fgithub.com\u002FBagelHole\u002FDevOps-Security-Agent-Skills","devops",{"author":46,"version":47},"devops-skills","1.0","2026-09-20","BagelHole\u002FDevOps-Security-Agent-Skills","Requires the relevant OS\u002Fplatform tooling and privileged access where noted. Docs-only; helper scripts and templates not bundled.","https:\u002F\u002Fgithub.com\u002FBagelHole\u002FDevOps-Security-Agent-Skills\u002Fblob\u002Fmain\u002FLICENSE","f3312ba749561b6159f477dbfc9313c875d40a56",".agents\u002Fskills_archive\u002Fllm-inference-scaling","https:\u002F\u002Fgithub.com\u002Fduc-duy-1612\u002FOnline-Delivery-Systems-Web","https:\u002F\u002Fgithub.com\u002Fduc-duy-1612","2026-10-25 23:50:58",{"source":8,"skillName":5,"commands":58,"prompt":68,"zipAvailable":69},[59,62,65],{"tool":60,"command":61},"npx","npx skills add duc-duy-1612\u002FOnline-Delivery-Systems-Web --skill llm-inference-scaling",{"tool":63,"command":64},"bunx","bunx skills add duc-duy-1612\u002FOnline-Delivery-Systems-Web --skill llm-inference-scaling",{"tool":66,"command":67},"pnpm","pnpm dlx skills add duc-duy-1612\u002FOnline-Delivery-Systems-Web --skill llm-inference-scaling","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fduc-duy-1612\u002FOnline-Delivery-Systems-Web\u002Fblob\u002Fmain\u002F.agents\u002Fskills_archive\u002Fllm-inference-scaling\u002FSKILL.md",true,[],[72,95,119,144,172,196],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":83,"publishedAt":38},"01M6Q0FA5FZ410Z33Z3TFQ4DBA","zero-trust","Implement zero-trust network architecture. Configure identity-based access, micro-segmentation, and continuous verification. Use when implementing modern security architectures.","https:\u002F\u002Fgithub.com\u002Fduc-duy-1612\u002FOnline-Delivery-Systems-Web\u002Fblob\u002Fmain\u002F.agents\u002Fskills_archive\u002Fzero-trust\u002FSKILL.md",[78],{"categoryNo":79,"categoryName":80,"taxonomyType":81,"taxonomyTypeDesc":82},"01JBSEEDCAT000000000000011","运维与部署","SCENARIO","使用场景",[84,85,86,87,88,89,74,90,35,34,91,32,31,22,92,27,93,26,25,94],"ZeRO","trust","zero trust","architecture","assessment","verification","segmentation","configure","active","scope","implementing",{"skillNo":96,"name":97,"nameZh":6,"description":98,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":99,"categories":100,"keywords":101,"publishedAt":38},"01M6Q0F1YZJW3KX5FBKEVZ8CV4","zcode-delegate","Delegate coding tasks to the Z.AI ZCode CLI only when the user explicitly requests it, while the orchestrator retains review and landing responsibility.","https:\u002F\u002Fgithub.com\u002Fduc-duy-1612\u002FOnline-Delivery-Systems-Web\u002Fblob\u002Fmain\u002F.agents\u002Fskills_archive\u002Fzcode-delegate\u002FSKILL.md",[],[102,103,104,105,106,107,108,109,110,111,112,113,114,32,115,31,116,117,27,118],"delegate","zcode","zcode delegate","review","cli","orchestration","tasks","run","Bash","model","node","read","orchestrator","files","desktop app","app","agent-orchestration",{"skillNo":120,"name":121,"nameZh":6,"description":122,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":123,"categories":124,"keywords":125,"publishedAt":38},"01M6Q0ETK41VGNMYEXW0T00AJN","youtube-transcript","Fetch YouTube transcripts through DeepAPI or local fallback tooling and save clean text output.","https:\u002F\u002Fgithub.com\u002Fduc-duy-1612\u002FOnline-Delivery-Systems-Web\u002Fblob\u002Fmain\u002F.agents\u002Fskills_archive\u002Fyoutube-transcript\u002FSKILL.md",[],[126,127,128,129,130,131,132,133,134,135,136,137,138,31,22,139,140,141,142,143],"transcript","youtube transcript","youtube","safe","github","codex","research","license","https","text","fetch","tooling","clean","local","output","deepapi","youtube transcripts","blob",{"skillNo":145,"name":146,"nameZh":6,"description":147,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":148,"categories":149,"keywords":161,"publishedAt":38},"01M6Q0EPPH2PVMDW40017F8WX0","youtube-transcript-skills","Fetch YouTube video transcripts, search videos\u002Fchannels, browse channels, and extract playlists via the getyoutubetranscript.com API - free tier, no card required.","https:\u002F\u002Fgithub.com\u002Fduc-duy-1612\u002FOnline-Delivery-Systems-Web\u002Fblob\u002Fmain\u002F.agents\u002Fskills_archive\u002Fyoutube-transcript-skills\u002FSKILL.md",[150,153,156],{"categoryNo":151,"categoryName":152,"taxonomyType":81,"taxonomyTypeDesc":82},"01JBSEEDCAT000000000000010","后端与接口",{"categoryNo":154,"categoryName":155,"taxonomyType":81,"taxonomyTypeDesc":82},"01JBSEEDCAT000000000000016","图像与多媒体",{"categoryNo":157,"categoryName":158,"taxonomyType":159,"taxonomyTypeDesc":160},"01JBSEEDCAT000000000000030","开发工程师","PROFESSION","适用职业",[126,127,128,162,129,163,130,131,133,164,165,166,167,136,168,31,22,169,170,171],"youtube transcript skills","gemini","search","api","Video","browse","api-integration","free","card","channels",{"skillNo":173,"name":174,"nameZh":6,"description":175,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":176,"categories":177,"keywords":182,"publishedAt":38},"01M6Q0EAB96PX83W5PTE8A0R8D","youtube-notetaker","Turn YouTube talks into local study notes with slides, transcripts, editable annotations, and a markdown-backed viewer.","https:\u002F\u002Fgithub.com\u002Fduc-duy-1612\u002FOnline-Delivery-Systems-Web\u002Fblob\u002Fmain\u002F.agents\u002Fskills_archive\u002Fyoutube-notetaker\u002FSKILL.md",[178,181],{"categoryNo":179,"categoryName":180,"taxonomyType":81,"taxonomyTypeDesc":82},"01JBSEEDCAT000000000000003","演示文稿",{"categoryNo":154,"categoryName":155,"taxonomyType":81,"taxonomyTypeDesc":82},[128,183,184,106,129,185,186,187,188,133,166,189,190,191,192,22,193,139,194,195],"youtube notetaker","notetaker","workflow","official","notes","code","markdown","slides","annotations","README","study","dair-academy","viewer",{"skillNo":197,"name":198,"nameZh":6,"description":199,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":200,"categories":201,"keywords":203,"publishedAt":38},"01M6Q0E55F0GW5MX48V8HXS8B5","youtube-full","Fetch YouTube transcripts, search videos, browse channels, and extract playlists via TranscriptAPI — no yt-dlp, no Google API key, works from any cloud server.","https:\u002F\u002Fgithub.com\u002Fduc-duy-1612\u002FOnline-Delivery-Systems-Web\u002Fblob\u002Fmain\u002F.agents\u002Fskills_archive\u002Fyoutube-full\u002FSKILL.md",[202],{"categoryNo":151,"categoryName":152,"taxonomyType":81,"taxonomyTypeDesc":82},[204,128,205,206,129,163,164,207,208,165,166,167,136,209,168,31,22,169,210,142],"youtube full","full","google","server","dlp","credits","automatic"]