[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-google-cloud-gke-inference-quickstart":3,"mdc-segqca-key":34,"related-org-google-cloud-gke-inference-quickstart":724,"related-repo-google-cloud-gke-inference-quickstart":913},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":12,"stars":24,"repoUrl":25,"updatedAt":26,"license":27,"forks":28,"topics":29,"repo":30,"sourceUrl":32,"mdContent":33},"gke-inference-quickstart","deploy AI inference workloads on GKE","Deploy optimized AI\u002FML inference workloads on GKE using Google's Inference Quickstart (GIQ). Covers model discovery, manifest generation, and deployment using native MCP tools and CLI.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},"google-cloud","Google Cloud","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fgoogle-cloud.png","GoogleCloudPlatform",[13,17,20,23],{"name":14,"slug":15,"type":16},"Machine Learning","machine-learning","tag",{"name":18,"slug":19,"type":16},"Deployment","deployment",{"name":21,"slug":22,"type":16},"Kubernetes","kubernetes",{"name":9,"slug":8,"type":16},161,"https:\u002F\u002Fgithub.com\u002FGoogleCloudPlatform\u002Fgke-mcp","2026-07-12T07:39:57.105576",null,78,[],{"repoUrl":25,"stars":24,"forks":28,"topics":31,"description":27},[],"https:\u002F\u002Fgithub.com\u002FGoogleCloudPlatform\u002Fgke-mcp\u002Ftree\u002FHEAD\u002Fskills\u002Fgke-inference-quickstart","---\nname: gke-inference-quickstart\ndescription: Deploy optimized AI\u002FML inference workloads on GKE using Google's Inference Quickstart (GIQ). Covers model discovery, manifest generation, and deployment using native MCP tools and CLI.\n---\n\n# GKE Inference Quickstart (GIQ)\n\n## Purpose\n\nThis skill guides the deployment of AI\u002FML inference workloads on GKE using GIQ. It leverages `gcloud container ai profiles manifests create` to create optimized Kubernetes manifests based on Google's best practices and benchmarks.\n\n## When to Use\n\n- **Goal:** Deploy an AI model (e.g., Llama, Gemma, Mistral) to GKE.\n- **Goal:** Generate a Kubernetes manifest for inference.\n- **Context:** User asks about \"GIQ\", \"Inference Quickstart\", or \"AI benchmarks\" on GKE.\n\n## Prerequisites\n\n- A GKE cluster (preferably with GPU\u002FTPU node pools, though GIQ can help identify requirements).\n- `gcloud` CLI installed and authenticated (for discovery commands).\n\n## Workflow\n\n### 1. Discovery: Find Models and Hardware\n\nBefore generating a manifest, you often need to pick a valid combination of Model, Model Server, and Accelerator.\n\n**List all supported models:**\n\n```bash\ngcloud container ai profiles models list\n```\n\n**Find valid accelerators and servers for a specific model:**\n\n```bash\n# Replace \u003CMODEL_NAME> with a model from the list above (e.g., 'gemma-2-9b-it')\ngcloud container ai profiles list --model=\u003CMODEL_NAME>\n```\n\n**View benchmarks\u002Fprofiles (optional):**\nTo see costs and latency targets:\n\n```bash\ngcloud container ai profiles list --model=\u003CMODEL_NAME>\n```\n\n### 2. Generate Manifest\n\nUse the `gcloud container ai profiles manifests create` command. This ensures you are using the latest supported flags and options directly from the CLI.\n\n**Parameters:**\n\n- `--model`: The model ID (e.g., `gemma-2-9b-it`).\n- `--model-server`: The inference server (e.g., `vllm`, `tgi`, `triton`, `tensorrt-llm`).\n- `--accelerator-type`: The accelerator type (e.g., `nvidia-l4`, `nvidia-tesla-a100`).\n- `--target-ntpot-milliseconds`: (Optional) Target Normalized Time Per Output Token in ms.\n\n**Example Command:**\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 > inference-workload.yaml\n```\n\n### 3. Review and Deploy\n\n1. **Save:** The example command above saves output to `inference-workload.yaml`. Ensure you have this file.\n2. **Review:** Check for any placeholders or specific requirements (like PVCs or secrets).\n   - _Note: Some models require Hugging Face tokens. Ensure query instructions for secrets are followed._\n3. **Deploy:**\n   ```bash\n   kubectl apply -f inference-workload.yaml\n   ```\n\n## Troubleshooting\n\n- **Invalid Combination:** If the manifest creation fails with an invalid combination error, re-run the discovery commands in Step 1 to verify the tuple (model, server, accelerator).\n- **Quota Issues:** Ensure the target region has sufficient quota for the requested accelerator (e.g., `NVIDIA_L4_GPUS`).\n\n## Reference\n\n- **Docs:** [GKE Inference Quickstart Documentation](https:\u002F\u002Fdocs.cloud.google.com\u002Fkubernetes-engine\u002Fdocs\u002Fhow-to\u002Fmachine-learning\u002Finference\u002Finference-quickstart)\n",{"data":35,"body":36},{"name":4,"description":6},{"type":37,"children":38},"root",[39,48,55,70,76,111,117,136,142,149,154,162,210,218,280,290,336,342,354,362,460,468,567,573,654,660,690,696,718],{"type":40,"tag":41,"props":42,"children":44},"element","h1",{"id":43},"gke-inference-quickstart-giq",[45],{"type":46,"value":47},"text","GKE Inference Quickstart (GIQ)",{"type":40,"tag":49,"props":50,"children":52},"h2",{"id":51},"purpose",[53],{"type":46,"value":54},"Purpose",{"type":40,"tag":56,"props":57,"children":58},"p",{},[59,61,68],{"type":46,"value":60},"This skill guides the deployment of AI\u002FML inference workloads on GKE using GIQ. It leverages ",{"type":40,"tag":62,"props":63,"children":65},"code",{"className":64},[],[66],{"type":46,"value":67},"gcloud container ai profiles manifests create",{"type":46,"value":69}," to create optimized Kubernetes manifests based on Google's best practices and benchmarks.",{"type":40,"tag":49,"props":71,"children":73},{"id":72},"when-to-use",[74],{"type":46,"value":75},"When to Use",{"type":40,"tag":77,"props":78,"children":79},"ul",{},[80,92,101],{"type":40,"tag":81,"props":82,"children":83},"li",{},[84,90],{"type":40,"tag":85,"props":86,"children":87},"strong",{},[88],{"type":46,"value":89},"Goal:",{"type":46,"value":91}," Deploy an AI model (e.g., Llama, Gemma, Mistral) to GKE.",{"type":40,"tag":81,"props":93,"children":94},{},[95,99],{"type":40,"tag":85,"props":96,"children":97},{},[98],{"type":46,"value":89},{"type":46,"value":100}," Generate a Kubernetes manifest for inference.",{"type":40,"tag":81,"props":102,"children":103},{},[104,109],{"type":40,"tag":85,"props":105,"children":106},{},[107],{"type":46,"value":108},"Context:",{"type":46,"value":110}," User asks about \"GIQ\", \"Inference Quickstart\", or \"AI benchmarks\" on GKE.",{"type":40,"tag":49,"props":112,"children":114},{"id":113},"prerequisites",[115],{"type":46,"value":116},"Prerequisites",{"type":40,"tag":77,"props":118,"children":119},{},[120,125],{"type":40,"tag":81,"props":121,"children":122},{},[123],{"type":46,"value":124},"A GKE cluster (preferably with GPU\u002FTPU node pools, though GIQ can help identify requirements).",{"type":40,"tag":81,"props":126,"children":127},{},[128,134],{"type":40,"tag":62,"props":129,"children":131},{"className":130},[],[132],{"type":46,"value":133},"gcloud",{"type":46,"value":135}," CLI installed and authenticated (for discovery commands).",{"type":40,"tag":49,"props":137,"children":139},{"id":138},"workflow",[140],{"type":46,"value":141},"Workflow",{"type":40,"tag":143,"props":144,"children":146},"h3",{"id":145},"_1-discovery-find-models-and-hardware",[147],{"type":46,"value":148},"1. Discovery: Find Models and Hardware",{"type":40,"tag":56,"props":150,"children":151},{},[152],{"type":46,"value":153},"Before generating a manifest, you often need to pick a valid combination of Model, Model Server, and Accelerator.",{"type":40,"tag":56,"props":155,"children":156},{},[157],{"type":40,"tag":85,"props":158,"children":159},{},[160],{"type":46,"value":161},"List all supported models:",{"type":40,"tag":163,"props":164,"children":169},"pre",{"className":165,"code":166,"language":167,"meta":168,"style":168},"language-bash shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","gcloud container ai profiles models list\n","bash","",[170],{"type":40,"tag":62,"props":171,"children":172},{"__ignoreMap":168},[173],{"type":40,"tag":174,"props":175,"children":178},"span",{"class":176,"line":177},"line",1,[179,184,190,195,200,205],{"type":40,"tag":174,"props":180,"children":182},{"style":181},"--shiki-light:#E2931D;--shiki-default:#FFCB6B;--shiki-dark:#FFCB6B",[183],{"type":46,"value":133},{"type":40,"tag":174,"props":185,"children":187},{"style":186},"--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D",[188],{"type":46,"value":189}," container",{"type":40,"tag":174,"props":191,"children":192},{"style":186},[193],{"type":46,"value":194}," ai",{"type":40,"tag":174,"props":196,"children":197},{"style":186},[198],{"type":46,"value":199}," profiles",{"type":40,"tag":174,"props":201,"children":202},{"style":186},[203],{"type":46,"value":204}," models",{"type":40,"tag":174,"props":206,"children":207},{"style":186},[208],{"type":46,"value":209}," list\n",{"type":40,"tag":56,"props":211,"children":212},{},[213],{"type":40,"tag":85,"props":214,"children":215},{},[216],{"type":46,"value":217},"Find valid accelerators and servers for a specific model:",{"type":40,"tag":163,"props":219,"children":221},{"className":165,"code":220,"language":167,"meta":168,"style":168},"# Replace \u003CMODEL_NAME> with a model from the list above (e.g., 'gemma-2-9b-it')\ngcloud container ai profiles list --model=\u003CMODEL_NAME>\n",[222],{"type":40,"tag":62,"props":223,"children":224},{"__ignoreMap":168},[225,234],{"type":40,"tag":174,"props":226,"children":227},{"class":176,"line":177},[228],{"type":40,"tag":174,"props":229,"children":231},{"style":230},"--shiki-light:#90A4AE;--shiki-light-font-style:italic;--shiki-default:#546E7A;--shiki-default-font-style:italic;--shiki-dark:#676E95;--shiki-dark-font-style:italic",[232],{"type":46,"value":233},"# Replace \u003CMODEL_NAME> with a model from the list above (e.g., 'gemma-2-9b-it')\n",{"type":40,"tag":174,"props":235,"children":237},{"class":176,"line":236},2,[238,242,246,250,254,259,264,270,275],{"type":40,"tag":174,"props":239,"children":240},{"style":181},[241],{"type":46,"value":133},{"type":40,"tag":174,"props":243,"children":244},{"style":186},[245],{"type":46,"value":189},{"type":40,"tag":174,"props":247,"children":248},{"style":186},[249],{"type":46,"value":194},{"type":40,"tag":174,"props":251,"children":252},{"style":186},[253],{"type":46,"value":199},{"type":40,"tag":174,"props":255,"children":256},{"style":186},[257],{"type":46,"value":258}," list",{"type":40,"tag":174,"props":260,"children":261},{"style":186},[262],{"type":46,"value":263}," --model=",{"type":40,"tag":174,"props":265,"children":267},{"style":266},"--shiki-light:#39ADB5;--shiki-default:#89DDFF;--shiki-dark:#89DDFF",[268],{"type":46,"value":269},"\u003C",{"type":40,"tag":174,"props":271,"children":272},{"style":186},[273],{"type":46,"value":274},"MODEL_NAME",{"type":40,"tag":174,"props":276,"children":277},{"style":266},[278],{"type":46,"value":279},">\n",{"type":40,"tag":56,"props":281,"children":282},{},[283,288],{"type":40,"tag":85,"props":284,"children":285},{},[286],{"type":46,"value":287},"View benchmarks\u002Fprofiles (optional):",{"type":46,"value":289},"\nTo see costs and latency targets:",{"type":40,"tag":163,"props":291,"children":293},{"className":165,"code":292,"language":167,"meta":168,"style":168},"gcloud container ai profiles list --model=\u003CMODEL_NAME>\n",[294],{"type":40,"tag":62,"props":295,"children":296},{"__ignoreMap":168},[297],{"type":40,"tag":174,"props":298,"children":299},{"class":176,"line":177},[300,304,308,312,316,320,324,328,332],{"type":40,"tag":174,"props":301,"children":302},{"style":181},[303],{"type":46,"value":133},{"type":40,"tag":174,"props":305,"children":306},{"style":186},[307],{"type":46,"value":189},{"type":40,"tag":174,"props":309,"children":310},{"style":186},[311],{"type":46,"value":194},{"type":40,"tag":174,"props":313,"children":314},{"style":186},[315],{"type":46,"value":199},{"type":40,"tag":174,"props":317,"children":318},{"style":186},[319],{"type":46,"value":258},{"type":40,"tag":174,"props":321,"children":322},{"style":186},[323],{"type":46,"value":263},{"type":40,"tag":174,"props":325,"children":326},{"style":266},[327],{"type":46,"value":269},{"type":40,"tag":174,"props":329,"children":330},{"style":186},[331],{"type":46,"value":274},{"type":40,"tag":174,"props":333,"children":334},{"style":266},[335],{"type":46,"value":279},{"type":40,"tag":143,"props":337,"children":339},{"id":338},"_2-generate-manifest",[340],{"type":46,"value":341},"2. Generate Manifest",{"type":40,"tag":56,"props":343,"children":344},{},[345,347,352],{"type":46,"value":346},"Use the ",{"type":40,"tag":62,"props":348,"children":350},{"className":349},[],[351],{"type":46,"value":67},{"type":46,"value":353}," command. This ensures you are using the latest supported flags and options directly from the CLI.",{"type":40,"tag":56,"props":355,"children":356},{},[357],{"type":40,"tag":85,"props":358,"children":359},{},[360],{"type":46,"value":361},"Parameters:",{"type":40,"tag":77,"props":363,"children":364},{},[365,384,424,449],{"type":40,"tag":81,"props":366,"children":367},{},[368,374,376,382],{"type":40,"tag":62,"props":369,"children":371},{"className":370},[],[372],{"type":46,"value":373},"--model",{"type":46,"value":375},": The model ID (e.g., ",{"type":40,"tag":62,"props":377,"children":379},{"className":378},[],[380],{"type":46,"value":381},"gemma-2-9b-it",{"type":46,"value":383},").",{"type":40,"tag":81,"props":385,"children":386},{},[387,393,395,401,403,409,410,416,417,423],{"type":40,"tag":62,"props":388,"children":390},{"className":389},[],[391],{"type":46,"value":392},"--model-server",{"type":46,"value":394},": The inference server (e.g., ",{"type":40,"tag":62,"props":396,"children":398},{"className":397},[],[399],{"type":46,"value":400},"vllm",{"type":46,"value":402},", ",{"type":40,"tag":62,"props":404,"children":406},{"className":405},[],[407],{"type":46,"value":408},"tgi",{"type":46,"value":402},{"type":40,"tag":62,"props":411,"children":413},{"className":412},[],[414],{"type":46,"value":415},"triton",{"type":46,"value":402},{"type":40,"tag":62,"props":418,"children":420},{"className":419},[],[421],{"type":46,"value":422},"tensorrt-llm",{"type":46,"value":383},{"type":40,"tag":81,"props":425,"children":426},{},[427,433,435,441,442,448],{"type":40,"tag":62,"props":428,"children":430},{"className":429},[],[431],{"type":46,"value":432},"--accelerator-type",{"type":46,"value":434},": The accelerator type (e.g., ",{"type":40,"tag":62,"props":436,"children":438},{"className":437},[],[439],{"type":46,"value":440},"nvidia-l4",{"type":46,"value":402},{"type":40,"tag":62,"props":443,"children":445},{"className":444},[],[446],{"type":46,"value":447},"nvidia-tesla-a100",{"type":46,"value":383},{"type":40,"tag":81,"props":450,"children":451},{},[452,458],{"type":40,"tag":62,"props":453,"children":455},{"className":454},[],[456],{"type":46,"value":457},"--target-ntpot-milliseconds",{"type":46,"value":459},": (Optional) Target Normalized Time Per Output Token in ms.",{"type":40,"tag":56,"props":461,"children":462},{},[463],{"type":40,"tag":85,"props":464,"children":465},{},[466],{"type":46,"value":467},"Example Command:",{"type":40,"tag":163,"props":469,"children":471},{"className":165,"code":470,"language":167,"meta":168,"style":168},"gcloud 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 > inference-workload.yaml\n",[472],{"type":40,"tag":62,"props":473,"children":474},{"__ignoreMap":168},[475,510,522,535,548],{"type":40,"tag":174,"props":476,"children":477},{"class":176,"line":177},[478,482,486,490,494,499,504],{"type":40,"tag":174,"props":479,"children":480},{"style":181},[481],{"type":46,"value":133},{"type":40,"tag":174,"props":483,"children":484},{"style":186},[485],{"type":46,"value":189},{"type":40,"tag":174,"props":487,"children":488},{"style":186},[489],{"type":46,"value":194},{"type":40,"tag":174,"props":491,"children":492},{"style":186},[493],{"type":46,"value":199},{"type":40,"tag":174,"props":495,"children":496},{"style":186},[497],{"type":46,"value":498}," manifests",{"type":40,"tag":174,"props":500,"children":501},{"style":186},[502],{"type":46,"value":503}," create",{"type":40,"tag":174,"props":505,"children":507},{"style":506},"--shiki-light:#90A4AE;--shiki-default:#EEFFFF;--shiki-dark:#BABED8",[508],{"type":46,"value":509}," \\\n",{"type":40,"tag":174,"props":511,"children":512},{"class":176,"line":236},[513,518],{"type":40,"tag":174,"props":514,"children":515},{"style":186},[516],{"type":46,"value":517},"  --model=gemma-2-9b-it",{"type":40,"tag":174,"props":519,"children":520},{"style":506},[521],{"type":46,"value":509},{"type":40,"tag":174,"props":523,"children":525},{"class":176,"line":524},3,[526,531],{"type":40,"tag":174,"props":527,"children":528},{"style":186},[529],{"type":46,"value":530},"  --model-server=vllm",{"type":40,"tag":174,"props":532,"children":533},{"style":506},[534],{"type":46,"value":509},{"type":40,"tag":174,"props":536,"children":538},{"class":176,"line":537},4,[539,544],{"type":40,"tag":174,"props":540,"children":541},{"style":186},[542],{"type":46,"value":543},"  --accelerator-type=nvidia-l4",{"type":40,"tag":174,"props":545,"children":546},{"style":506},[547],{"type":46,"value":509},{"type":40,"tag":174,"props":549,"children":551},{"class":176,"line":550},5,[552,557,562],{"type":40,"tag":174,"props":553,"children":554},{"style":186},[555],{"type":46,"value":556},"  --target-ntpot-milliseconds=50",{"type":40,"tag":174,"props":558,"children":559},{"style":266},[560],{"type":46,"value":561}," >",{"type":40,"tag":174,"props":563,"children":564},{"style":186},[565],{"type":46,"value":566}," inference-workload.yaml\n",{"type":40,"tag":143,"props":568,"children":570},{"id":569},"_3-review-and-deploy",[571],{"type":46,"value":572},"3. Review and Deploy",{"type":40,"tag":574,"props":575,"children":576},"ol",{},[577,595,617],{"type":40,"tag":81,"props":578,"children":579},{},[580,585,587,593],{"type":40,"tag":85,"props":581,"children":582},{},[583],{"type":46,"value":584},"Save:",{"type":46,"value":586}," The example command above saves output to ",{"type":40,"tag":62,"props":588,"children":590},{"className":589},[],[591],{"type":46,"value":592},"inference-workload.yaml",{"type":46,"value":594},". Ensure you have this file.",{"type":40,"tag":81,"props":596,"children":597},{},[598,603,605],{"type":40,"tag":85,"props":599,"children":600},{},[601],{"type":46,"value":602},"Review:",{"type":46,"value":604}," Check for any placeholders or specific requirements (like PVCs or secrets).\n",{"type":40,"tag":77,"props":606,"children":607},{},[608],{"type":40,"tag":81,"props":609,"children":610},{},[611],{"type":40,"tag":612,"props":613,"children":614},"em",{},[615],{"type":46,"value":616},"Note: Some models require Hugging Face tokens. Ensure query instructions for secrets are followed.",{"type":40,"tag":81,"props":618,"children":619},{},[620,625],{"type":40,"tag":85,"props":621,"children":622},{},[623],{"type":46,"value":624},"Deploy:",{"type":40,"tag":163,"props":626,"children":628},{"className":165,"code":627,"language":167,"meta":168,"style":168},"kubectl apply -f inference-workload.yaml\n",[629],{"type":40,"tag":62,"props":630,"children":631},{"__ignoreMap":168},[632],{"type":40,"tag":174,"props":633,"children":634},{"class":176,"line":177},[635,640,645,650],{"type":40,"tag":174,"props":636,"children":637},{"style":181},[638],{"type":46,"value":639},"kubectl",{"type":40,"tag":174,"props":641,"children":642},{"style":186},[643],{"type":46,"value":644}," apply",{"type":40,"tag":174,"props":646,"children":647},{"style":186},[648],{"type":46,"value":649}," -f",{"type":40,"tag":174,"props":651,"children":652},{"style":186},[653],{"type":46,"value":566},{"type":40,"tag":49,"props":655,"children":657},{"id":656},"troubleshooting",[658],{"type":46,"value":659},"Troubleshooting",{"type":40,"tag":77,"props":661,"children":662},{},[663,673],{"type":40,"tag":81,"props":664,"children":665},{},[666,671],{"type":40,"tag":85,"props":667,"children":668},{},[669],{"type":46,"value":670},"Invalid Combination:",{"type":46,"value":672}," If the manifest creation fails with an invalid combination error, re-run the discovery commands in Step 1 to verify the tuple (model, server, accelerator).",{"type":40,"tag":81,"props":674,"children":675},{},[676,681,683,689],{"type":40,"tag":85,"props":677,"children":678},{},[679],{"type":46,"value":680},"Quota Issues:",{"type":46,"value":682}," Ensure the target region has sufficient quota for the requested accelerator (e.g., ",{"type":40,"tag":62,"props":684,"children":686},{"className":685},[],[687],{"type":46,"value":688},"NVIDIA_L4_GPUS",{"type":46,"value":383},{"type":40,"tag":49,"props":691,"children":693},{"id":692},"reference",[694],{"type":46,"value":695},"Reference",{"type":40,"tag":77,"props":697,"children":698},{},[699],{"type":40,"tag":81,"props":700,"children":701},{},[702,707,709],{"type":40,"tag":85,"props":703,"children":704},{},[705],{"type":46,"value":706},"Docs:",{"type":46,"value":708}," ",{"type":40,"tag":710,"props":711,"children":715},"a",{"href":712,"rel":713},"https:\u002F\u002Fdocs.cloud.google.com\u002Fkubernetes-engine\u002Fdocs\u002Fhow-to\u002Fmachine-learning\u002Finference\u002Finference-quickstart",[714],"nofollow",[716],{"type":46,"value":717},"GKE Inference Quickstart Documentation",{"type":40,"tag":719,"props":720,"children":721},"style",{},[722],{"type":46,"value":723},"html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"items":725,"total":912},[726,744,760,782,796,807,821,838,855,868,884,894],{"slug":727,"name":727,"fn":728,"description":729,"org":730,"tags":731,"stars":741,"repoUrl":742,"updatedAt":743},"kb-search","search and extract local knowledge base documents","Allows listing, searching and extracting information from local knowledge base documents for information about tables\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[732,735,738],{"name":733,"slug":734,"type":16},"Documentation","documentation",{"name":736,"slug":737,"type":16},"Knowledge Base","knowledge-base",{"name":739,"slug":740,"type":16},"Search","search",6749,"https:\u002F\u002Fgithub.com\u002FGoogleCloudPlatform\u002Fknowledge-catalog","2026-07-12T07:38:52.157375",{"slug":745,"name":746,"fn":747,"description":748,"org":749,"tags":750,"stars":741,"repoUrl":742,"updatedAt":759},"knowledgecatalogdiscoveryagent","knowledge_catalog_discovery_agent","search and rank Knowledge Catalog data entries","Analyzes user queries, extracts relevant predicates, and utilizes Knowledge Catalog Search to find and rank the most relevant data entries. Engages with the user throughout the process.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[751,754,755,758],{"name":752,"slug":753,"type":16},"Data Analysis","data-analysis",{"name":9,"slug":8,"type":16},{"name":756,"slug":757,"type":16},"Knowledge Management","knowledge-management",{"name":739,"slug":740,"type":16},"2026-07-12T07:38:22.196851",{"slug":761,"name":761,"fn":762,"description":763,"org":764,"tags":765,"stars":779,"repoUrl":780,"updatedAt":781},"contributing","contribute to Cloud Foundation Fabric","End-to-end workflow for contributing to Cloud Foundation Fabric: triaging GitHub issues, proactive feature development, validating with tests and Policy Troubleshooter, and submitting sanitized Pull Requests. Use when addressing a Fabric GitHub issue, developing a module or FAST stage change, or preparing a branch for a pull request.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[766,769,772,775,776],{"name":767,"slug":768,"type":16},"Automation","automation",{"name":770,"slug":771,"type":16},"Engineering","engineering",{"name":773,"slug":774,"type":16},"GitHub","github",{"name":9,"slug":8,"type":16},{"name":777,"slug":778,"type":16},"Pull Requests","pull-requests",2062,"https:\u002F\u002Fgithub.com\u002FGoogleCloudPlatform\u002Fcloud-foundation-fabric","2026-07-31T06:23:36.935005",{"slug":783,"name":783,"fn":784,"description":785,"org":786,"tags":787,"stars":779,"repoUrl":780,"updatedAt":795},"fabric-builder","generate Terraform code for Google Cloud","Generates idiomatic Cloud Foundation Fabric (CFF) Terraform code using CFF modules. Use when users ask to create GCP resources, use Fabric modules, or generate Terraform code for Google Cloud.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[788,789,792],{"name":9,"slug":8,"type":16},{"name":790,"slug":791,"type":16},"Infrastructure as Code","infrastructure-as-code",{"name":793,"slug":794,"type":16},"Terraform","terraform","2026-07-12T07:38:23.514555",{"slug":797,"name":797,"fn":798,"description":799,"org":800,"tags":801,"stars":779,"repoUrl":780,"updatedAt":806},"fast-0-org-setup-prereqs","prepare prerequisites for FAST 0-org-setup","Guides the user step-by-step through the prerequisites for the FAST 0-org-setup stage, supporting both Standard GCP and Google Cloud Dedicated (GCD) environments. Use when a user asks to prepare or run prerequisites for 0-org-setup or bootstrap the FAST landing zone.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[802,803],{"name":9,"slug":8,"type":16},{"name":804,"slug":805,"type":16},"Operations","operations","2026-07-12T07:38:28.127148",{"slug":808,"name":808,"fn":809,"description":810,"org":811,"tags":812,"stars":818,"repoUrl":819,"updatedAt":820},"agent-aware-cli","design agent-aware command-line interfaces","Guide for designing and implementing command-line interfaces (CLIs) that are equally usable by human developers and automated coding agents. Use when the user wants to build a CLI, apply CLI best practices, or use Go with Cobra and Viper.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[813,816,817],{"name":814,"slug":815,"type":16},"CLI","cli",{"name":770,"slug":771,"type":16},{"name":9,"slug":8,"type":16},1150,"https:\u002F\u002Fgithub.com\u002FGoogleCloudPlatform\u002Fvertex-ai-creative-studio","2026-07-12T07:39:08.41406",{"slug":822,"name":822,"fn":823,"description":824,"org":825,"tags":826,"stars":818,"repoUrl":819,"updatedAt":837},"build-mcp-genmedia","build and configure GenAI MCP servers","Builds the mcp-genmedia Go MCP servers (nanobanana, veo, lyria, gemini-multimodal, chirp3-hd, avtool) from source and wires them into settings.json. Use this skill whenever the MCP tools are missing or broken — typically at the start of a new session, after a container restart, or when \u002Ftmp has been wiped. The prebuilt binaries in \u002Fworkspace\u002F.local\u002Fbin\u002F have no exec bit and live on a noexec mount; this skill compiles fresh executables into \u002Ftmp\u002Fbin\u002F where execution is allowed.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[827,830,831,834],{"name":828,"slug":829,"type":16},"API Development","api-development",{"name":9,"slug":8,"type":16},{"name":832,"slug":833,"type":16},"LLM","llm",{"name":835,"slug":836,"type":16},"MCP","mcp","2026-07-12T07:39:10.911302",{"slug":839,"name":839,"fn":840,"description":841,"org":842,"tags":843,"stars":818,"repoUrl":819,"updatedAt":854},"genmedia-audio-engineer","synthesize and mix audio content","Expert in audio synthesis, music generation, and mixing. Use when creating podcasts, background scores, or multi-track audio layering using mcp-chirp3-go, mcp-lyria-go, mcp-gemini-go, mcp-nanobanana-go, and mcp-avtool-go.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[844,847,850,851],{"name":845,"slug":846,"type":16},"Audio","audio",{"name":848,"slug":849,"type":16},"Creative","creative",{"name":9,"slug":8,"type":16},{"name":852,"slug":853,"type":16},"Vertex AI","vertex-ai","2026-07-12T07:39:16.623879",{"slug":856,"name":856,"fn":857,"description":858,"org":859,"tags":860,"stars":818,"repoUrl":819,"updatedAt":867},"genmedia-image-artist","generate and edit AI images","Expert in AI image generation and editing. Use when the user needs high-quality textures, character-consistent visuals, or image-to-image editing using mcp-nanobanana-go.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[861,862,863,866],{"name":848,"slug":849,"type":16},{"name":9,"slug":8,"type":16},{"name":864,"slug":865,"type":16},"Image Generation","image-generation",{"name":852,"slug":853,"type":16},"2026-07-12T07:39:15.372822",{"slug":869,"name":869,"fn":870,"description":871,"org":872,"tags":873,"stars":818,"repoUrl":819,"updatedAt":883},"genmedia-producer","produce multi-step media content","Expert media production assistant. Use when requested to help with storyboarding, podcast creation, audio assembly, or complex multi-step media workflows using the GenMedia MCP servers (Veo, Lyria, Gemini TTS, NanoBanana).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[874,875,876,877,880],{"name":845,"slug":846,"type":16},{"name":848,"slug":849,"type":16},{"name":9,"slug":8,"type":16},{"name":878,"slug":879,"type":16},"Media","media",{"name":881,"slug":882,"type":16},"Video","video","2026-07-12T07:39:09.672849",{"slug":885,"name":885,"fn":886,"description":887,"org":888,"tags":889,"stars":818,"repoUrl":819,"updatedAt":893},"genmedia-video-editor","edit and compose video content","Expert in video composition, editing, and format conversion. Use when the user wants to generate high-quality video, overlay images on video, concatenate clips, create GIFs, or sync audio to video using mcp-avtool-go and mcp-veo-go.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[890,891,892],{"name":848,"slug":849,"type":16},{"name":9,"slug":8,"type":16},{"name":881,"slug":882,"type":16},"2026-07-12T07:39:13.749081",{"slug":895,"name":895,"fn":896,"description":897,"org":898,"tags":899,"stars":818,"repoUrl":819,"updatedAt":911},"genmedia-voice-director","generate expressive text-to-speech with Gemini","Expert in casting, directing, and generating expressive text-to-speech using Gemini TTS. Use this when the user needs virtual voice actor personas, expressive speech generation, or multiple variations of a voiceover (like \"take 3 on the bounce\").",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[900,901,902,905,908],{"name":845,"slug":846,"type":16},{"name":848,"slug":849,"type":16},{"name":903,"slug":904,"type":16},"Gemini","gemini",{"name":906,"slug":907,"type":16},"Speech","speech",{"name":909,"slug":910,"type":16},"Text-to-Speech","text-to-speech","2026-07-12T07:39:17.86673",80,{"items":914,"total":1004},[915,925,940,953,966,977,993],{"slug":916,"name":916,"fn":917,"description":918,"org":919,"tags":920,"stars":24,"repoUrl":25,"updatedAt":924},"custom-golden-image-discovery","discover golden base images for GKE nodes","Expert at discovering golden base images for GKE custom nodes using technical specs or context clues.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[921,922,923],{"name":18,"slug":19,"type":16},{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},"2026-07-12T07:39:30.888879",{"slug":926,"name":926,"fn":927,"description":928,"org":929,"tags":930,"stars":24,"repoUrl":25,"updatedAt":939},"gke-ai-troubleshooting-handle-disruption-gpu-tpu","diagnose GPU and TPU workload disruptions","Diagnose and predict node disruption during Compute Engine host maintenance for GPU and TPU workloads.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[931,934,935,936],{"name":932,"slug":933,"type":16},"Debugging","debugging",{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":937,"slug":938,"type":16},"Performance","performance","2026-07-28T05:34:18.149515",{"slug":941,"name":941,"fn":942,"description":943,"org":944,"tags":945,"stars":24,"repoUrl":25,"updatedAt":952},"gke-ai-troubleshooting-jobset-interruption","diagnose GKE JobSet interruptions","Systematically diagnose GKE JobSet interruptions, restarts, and preemptions for AI\u002FML training workloads. Identifies preemption events, maintenance interruptions, bad host VMs, unhealthy pods, and coordinator worker failures.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[946,947,948,949],{"name":932,"slug":933,"type":16},{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":950,"slug":951,"type":16},"Observability","observability","2026-07-12T07:40:04.511878",{"slug":954,"name":954,"fn":955,"description":956,"org":957,"tags":958,"stars":24,"repoUrl":25,"updatedAt":965},"gke-ai-troubleshooting-skill-creation-guide","create GKE troubleshooting skill bundles","Expert instructions for building high-quality GKE troubleshooting skills. Codifies Step 0 context rules, zero-hallucination signatures, and explicit LQL\u002FPromQL query requirements.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[959,960,961,962],{"name":733,"slug":734,"type":16},{"name":770,"slug":771,"type":16},{"name":9,"slug":8,"type":16},{"name":963,"slug":964,"type":16},"Technical Writing","technical-writing","2026-07-12T07:39:50.73484",{"slug":967,"name":967,"fn":968,"description":969,"org":970,"tags":971,"stars":24,"repoUrl":25,"updatedAt":976},"gke-ai-troubleshooting-tpu-connection-failure-vbar-oom","diagnose GKE TPU connection failures","Diagnose and prevent `vbar_control_agent` segfaults and OOMs caused by race conditions during TPU device resets and frequent metrics collection (e.g. every 3s). Use when TPU slice initialization fails or `vbar_control_agent` crashes on TPU v6e nodes.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[972,973,974,975],{"name":932,"slug":933,"type":16},{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":937,"slug":938,"type":16},"2026-07-12T07:39:49.482979",{"slug":978,"name":978,"fn":979,"description":980,"org":981,"tags":982,"stars":24,"repoUrl":25,"updatedAt":992},"gke-app-onboarding","containerize and deploy apps to GKE","Workflows for containerizing and deploying applications to GKE for the first time.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[983,986,987,988,989],{"name":984,"slug":985,"type":16},"Containers","containers",{"name":18,"slug":19,"type":16},{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":990,"slug":991,"type":16},"Onboarding","onboarding","2026-07-12T07:39:41.935837",{"slug":994,"name":994,"fn":995,"description":996,"org":997,"tags":998,"stars":24,"repoUrl":25,"updatedAt":1003},"gke-backup-dr","configure GKE backup and disaster recovery","Workflows for configuring Backup for GKE and disaster recovery.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[999,1000,1001,1002],{"name":18,"slug":19,"type":16},{"name":9,"slug":8,"type":16},{"name":21,"slug":22,"type":16},{"name":804,"slug":805,"type":16},"2026-07-12T07:39:34.806995",25]