[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6Q02R5RYENT8WZ0GHS9P34Y":3,"skill-same:01M6Q02R5RYENT8WZ0GHS9P34Y":79},{"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":25,"publishedAt":46,"linkOnly":47,"markdownContent":48,"markdownBody":49,"frontMatter":50,"refName":29,"commitSha":60,"skillDirectory":61,"forksCount":13,"repositoryHtmlUrl":62,"repositoryHomepageUrl":6,"ownerHtmlUrl":63,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":64,"install":65,"resources":78},"01M6Q02R5RYENT8WZ0GHS9P34Y","vllm-server",null,"Deploy and manage vLLM for high-throughput LLM inference. Configure continuous batching, tensor parallelism, quantization, and OpenAI-compatible API endpoints for production LLM serving.","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\u002Fvllm-server\u002FSKILL.md",[17,22],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000010","后端与接口","SCENARIO","使用场景",{"categoryNo":23,"categoryName":24,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000011","运维与部署",[26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45],"server","vLLM","vllm server","main","for","openai-compatible","throughput","relevant","scripts","critical","os","platform","security","community","only","configure","access","api","devops-security","openai","2026-11-10 10:12:30",false,"---\nname: vllm-server\ndescription: Deploy and manage vLLM for high-throughput LLM inference. Configure continuous\n  batching, tensor parallelism, quantization, and OpenAI-compatible API endpoints\n  for production LLM serving.\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# vLLM Server Management\n\nDeploy production-grade LLM inference servers with vLLM — the fastest open-source LLM serving engine with PagedAttention and continuous batching.\n\n## When to Use This Skill\n\nUse this skill when:\n- Serving open-source LLMs (Llama, Mistral, Qwen, Gemma) at scale\n- Building an OpenAI-compatible API endpoint for self-hosted models\n- Optimizing LLM throughput and latency for production traffic\n- Running multi-GPU inference with tensor or pipeline parallelism\n- Deploying quantized models to reduce GPU memory requirements\n\n## Prerequisites\n\n- NVIDIA GPU(s) with CUDA 12.1+ (A100\u002FH100 recommended for production)\n- Docker or Python 3.9+ with pip\n- 40GB+ VRAM for 70B models; 8GB+ for 7B models\n- `nvidia-container-toolkit` for Docker GPU passthrough\n\n## Quick Start\n\n```bash\n# Install vLLM\npip install vllm\n\n# Serve a model (OpenAI-compatible API)\nvllm serve meta-llama\u002FLlama-3.1-8B-Instruct \\\n  --host 0.0.0.0 \\\n  --port 8000 \\\n  --api-key your-secret-key\n\n# Test the endpoint\ncurl http:\u002F\u002Flocalhost:8000\u002Fv1\u002Fchat\u002Fcompletions \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -H \"Authorization: Bearer your-secret-key\" \\\n  -d '{\n    \"model\": \"meta-llama\u002FLlama-3.1-8B-Instruct\",\n    \"messages\": [{\"role\": \"user\", \"content\": \"Hello!\"}]\n  }'\n```\n\n## Docker Deployment\n\n```bash\ndocker run --runtime nvidia --gpus all \\\n  -v ~\u002F.cache\u002Fhuggingface:\u002Froot\u002F.cache\u002Fhuggingface \\\n  -p 8000:8000 \\\n  --ipc=host \\\n  vllm\u002Fvllm-openai:latest \\\n  --model meta-llama\u002FLlama-3.1-8B-Instruct \\\n  --api-key your-secret-key\n```\n\n## Docker Compose (Production)\n\n```yaml\nservices:\n  vllm:\n    image: vllm\u002Fvllm-openai:latest\n    runtime: nvidia\n    environment:\n      - NVIDIA_VISIBLE_DEVICES=all\n      - HUGGING_FACE_HUB_TOKEN=${HF_TOKEN}\n    volumes:\n      - model-cache:\u002Froot\u002F.cache\u002Fhuggingface\n    ports:\n      - \"8000:8000\"\n    ipc: host\n    command: >\n      --model meta-llama\u002FLlama-3.1-70B-Instruct\n      --tensor-parallel-size 2\n      --max-model-len 32768\n      --gpu-memory-utilization 0.90\n      --api-key ${VLLM_API_KEY}\n    restart: unless-stopped\n    healthcheck:\n      test: [\"CMD\", \"curl\", \"-f\", \"http:\u002F\u002Flocalhost:8000\u002Fhealth\"]\n      interval: 30s\n      timeout: 10s\n      retries: 3\n\nvolumes:\n  model-cache:\n```\n\n## Key Configuration Options\n\n### Multi-GPU Tensor Parallelism\n\n```bash\n# Split one model across 4 GPUs\nvllm serve meta-llama\u002FLlama-3.1-70B-Instruct \\\n  --tensor-parallel-size 4 \\\n  --gpu-memory-utilization 0.90\n```\n\n### Quantization (Lower VRAM)\n\n```bash\n# AWQ quantization (70B on 2x A100 40GB)\nvllm serve casperhansen\u002Fllama-3-70b-instruct-awq \\\n  --quantization awq \\\n  --tensor-parallel-size 2\n\n# GPTQ quantization\nvllm serve TheBloke\u002FLlama-2-70B-Chat-GPTQ \\\n  --quantization gptq\n\n# FP8 (H100 NVL native)\nvllm serve meta-llama\u002FLlama-3.1-405B-Instruct \\\n  --quantization fp8 \\\n  --tensor-parallel-size 8\n```\n\n### Structured Output & Tools\n\n```bash\nvllm serve meta-llama\u002FLlama-3.1-8B-Instruct \\\n  --enable-auto-tool-choice \\\n  --tool-call-parser llama3_json \\\n  --guided-decoding-backend outlines\n```\n\n### LoRA Adapters\n\n```bash\nvllm serve meta-llama\u002FLlama-3.1-8B-Instruct \\\n  --enable-lora \\\n  --lora-modules sql-lora=\u002Fpath\u002Fto\u002Fsql-lora \\\n                 code-lora=\u002Fpath\u002Fto\u002Fcode-lora \\\n  --max-lora-rank 64\n```\n\n## Performance Tuning\n\n```bash\n# Maximize throughput for batch workloads\nvllm serve \u003Cmodel> \\\n  --max-num-seqs 256 \\          # max concurrent sequences\n  --max-num-batched-tokens 8192 \\ # tokens per batch\n  --gpu-memory-utilization 0.95 \\ # use 95% VRAM\n  --swap-space 4                  # CPU swap (GiB)\n\n# Minimize latency for interactive use\nvllm serve \u003Cmodel> \\\n  --max-num-seqs 32 \\\n  --enforce-eager              # disable CUDA graph capture\n```\n\n## Benchmarking\n\n```bash\n# Install benchmark tool\npip install vllm\n\n# Run throughput benchmark\npython -m vllm.entrypoints.openai.run_batch \\\n  --model meta-llama\u002FLlama-3.1-8B-Instruct \\\n  --input-file prompts.jsonl \\\n  --output-file results.jsonl\n\n# Benchmark with vllm bench\nvllm bench throughput \\\n  --model meta-llama\u002FLlama-3.1-8B-Instruct \\\n  --num-prompts 1000 \\\n  --input-len 512 \\\n  --output-len 128\n```\n\n## Monitoring\n\n```bash\n# Check running server stats\ncurl http:\u002F\u002Flocalhost:8000\u002Fmetrics  # Prometheus metrics\n\n# Key metrics to watch:\n# vllm:num_requests_running       - active requests\n# vllm:gpu_cache_usage_perc       - KV cache utilization\n# vllm:generation_tokens_per_s    - throughput\n# vllm:time_to_first_token_ms     - TTFT latency\n# vllm:e2e_request_latency_seconds - end-to-end latency\n```\n\n## Common Issues\n\n| Issue | Cause | Fix |\n|-------|-------|-----|\n| `CUDA out of memory` | Model too large for VRAM | Add `--quantization awq` or reduce `--gpu-memory-utilization` |\n| Slow cold start | Model not cached | Pre-pull with `huggingface-cli download \u003Cmodel>` |\n| Low throughput | Too few concurrent requests | Increase `--max-num-seqs` |\n| KV cache full errors | Context length too long | Set `--max-model-len` lower |\n| `tokenizer error` | Tokenizer mismatch | Use `--tokenizer` to specify correct tokenizer |\n\n## Best Practices\n\n- Use `--gpu-memory-utilization 0.90` to leave headroom for CUDA kernels.\n- Pin model versions with `--revision` for reproducible deployments.\n- Set `HF_HUB_OFFLINE=1` in production to prevent unexpected downloads.\n- Use AWQ or GPTQ quantization before tensor parallelism — lower VRAM first.\n- Enable `--enable-chunked-prefill` for long-context workloads.\n- Monitor `gpu_cache_usage_perc` — above 95% causes queuing.\n\n## Related Skills\n\n- llm-inference-scaling (`llm-inference-scaling`) - Auto-scaling vLLM deployments\n- gpu-server-management (`gpu-server-management`) - GPU driver setup\n- llm-gateway (`llm-gateway`) - Load balancing across vLLM instances\n- llm-cost-optimization (`llm-cost-optimization`) - Cost management\n- model-serving-kubernetes (`model-serving-kubernetes`) - K8s deployment\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","# vLLM Server Management\n\nDeploy production-grade LLM inference servers with vLLM — the fastest open-source LLM serving engine with PagedAttention and continuous batching.\n\n## When to Use This Skill\n\nUse this skill when:\n- Serving open-source LLMs (Llama, Mistral, Qwen, Gemma) at scale\n- Building an OpenAI-compatible API endpoint for self-hosted models\n- Optimizing LLM throughput and latency for production traffic\n- Running multi-GPU inference with tensor or pipeline parallelism\n- Deploying quantized models to reduce GPU memory requirements\n\n## Prerequisites\n\n- NVIDIA GPU(s) with CUDA 12.1+ (A100\u002FH100 recommended for production)\n- Docker or Python 3.9+ with pip\n- 40GB+ VRAM for 70B models; 8GB+ for 7B models\n- `nvidia-container-toolkit` for Docker GPU passthrough\n\n## Quick Start\n\n```bash\n# Install vLLM\npip install vllm\n\n# Serve a model (OpenAI-compatible API)\nvllm serve meta-llama\u002FLlama-3.1-8B-Instruct \\\n  --host 0.0.0.0 \\\n  --port 8000 \\\n  --api-key your-secret-key\n\n# Test the endpoint\ncurl http:\u002F\u002Flocalhost:8000\u002Fv1\u002Fchat\u002Fcompletions \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -H \"Authorization: Bearer your-secret-key\" \\\n  -d '{\n    \"model\": \"meta-llama\u002FLlama-3.1-8B-Instruct\",\n    \"messages\": [{\"role\": \"user\", \"content\": \"Hello!\"}]\n  }'\n```\n\n## Docker Deployment\n\n```bash\ndocker run --runtime nvidia --gpus all \\\n  -v ~\u002F.cache\u002Fhuggingface:\u002Froot\u002F.cache\u002Fhuggingface \\\n  -p 8000:8000 \\\n  --ipc=host \\\n  vllm\u002Fvllm-openai:latest \\\n  --model meta-llama\u002FLlama-3.1-8B-Instruct \\\n  --api-key your-secret-key\n```\n\n## Docker Compose (Production)\n\n```yaml\nservices:\n  vllm:\n    image: vllm\u002Fvllm-openai:latest\n    runtime: nvidia\n    environment:\n      - NVIDIA_VISIBLE_DEVICES=all\n      - HUGGING_FACE_HUB_TOKEN=${HF_TOKEN}\n    volumes:\n      - model-cache:\u002Froot\u002F.cache\u002Fhuggingface\n    ports:\n      - \"8000:8000\"\n    ipc: host\n    command: >\n      --model meta-llama\u002FLlama-3.1-70B-Instruct\n      --tensor-parallel-size 2\n      --max-model-len 32768\n      --gpu-memory-utilization 0.90\n      --api-key ${VLLM_API_KEY}\n    restart: unless-stopped\n    healthcheck:\n      test: [\"CMD\", \"curl\", \"-f\", \"http:\u002F\u002Flocalhost:8000\u002Fhealth\"]\n      interval: 30s\n      timeout: 10s\n      retries: 3\n\nvolumes:\n  model-cache:\n```\n\n## Key Configuration Options\n\n### Multi-GPU Tensor Parallelism\n\n```bash\n# Split one model across 4 GPUs\nvllm serve meta-llama\u002FLlama-3.1-70B-Instruct \\\n  --tensor-parallel-size 4 \\\n  --gpu-memory-utilization 0.90\n```\n\n### Quantization (Lower VRAM)\n\n```bash\n# AWQ quantization (70B on 2x A100 40GB)\nvllm serve casperhansen\u002Fllama-3-70b-instruct-awq \\\n  --quantization awq \\\n  --tensor-parallel-size 2\n\n# GPTQ quantization\nvllm serve TheBloke\u002FLlama-2-70B-Chat-GPTQ \\\n  --quantization gptq\n\n# FP8 (H100 NVL native)\nvllm serve meta-llama\u002FLlama-3.1-405B-Instruct \\\n  --quantization fp8 \\\n  --tensor-parallel-size 8\n```\n\n### Structured Output & Tools\n\n```bash\nvllm serve meta-llama\u002FLlama-3.1-8B-Instruct \\\n  --enable-auto-tool-choice \\\n  --tool-call-parser llama3_json \\\n  --guided-decoding-backend outlines\n```\n\n### LoRA Adapters\n\n```bash\nvllm serve meta-llama\u002FLlama-3.1-8B-Instruct \\\n  --enable-lora \\\n  --lora-modules sql-lora=\u002Fpath\u002Fto\u002Fsql-lora \\\n                 code-lora=\u002Fpath\u002Fto\u002Fcode-lora \\\n  --max-lora-rank 64\n```\n\n## Performance Tuning\n\n```bash\n# Maximize throughput for batch workloads\nvllm serve \u003Cmodel> \\\n  --max-num-seqs 256 \\          # max concurrent sequences\n  --max-num-batched-tokens 8192 \\ # tokens per batch\n  --gpu-memory-utilization 0.95 \\ # use 95% VRAM\n  --swap-space 4                  # CPU swap (GiB)\n\n# Minimize latency for interactive use\nvllm serve \u003Cmodel> \\\n  --max-num-seqs 32 \\\n  --enforce-eager              # disable CUDA graph capture\n```\n\n## Benchmarking\n\n```bash\n# Install benchmark tool\npip install vllm\n\n# Run throughput benchmark\npython -m vllm.entrypoints.openai.run_batch \\\n  --model meta-llama\u002FLlama-3.1-8B-Instruct \\\n  --input-file prompts.jsonl \\\n  --output-file results.jsonl\n\n# Benchmark with vllm bench\nvllm bench throughput \\\n  --model meta-llama\u002FLlama-3.1-8B-Instruct \\\n  --num-prompts 1000 \\\n  --input-len 512 \\\n  --output-len 128\n```\n\n## Monitoring\n\n```bash\n# Check running server stats\ncurl http:\u002F\u002Flocalhost:8000\u002Fmetrics  # Prometheus metrics\n\n# Key metrics to watch:\n# vllm:num_requests_running       - active requests\n# vllm:gpu_cache_usage_perc       - KV cache utilization\n# vllm:generation_tokens_per_s    - throughput\n# vllm:time_to_first_token_ms     - TTFT latency\n# vllm:e2e_request_latency_seconds - end-to-end latency\n```\n\n## Common Issues\n\n| Issue | Cause | Fix |\n|-------|-------|-----|\n| `CUDA out of memory` | Model too large for VRAM | Add `--quantization awq` or reduce `--gpu-memory-utilization` |\n| Slow cold start | Model not cached | Pre-pull with `huggingface-cli download \u003Cmodel>` |\n| Low throughput | Too few concurrent requests | Increase `--max-num-seqs` |\n| KV cache full errors | Context length too long | Set `--max-model-len` lower |\n| `tokenizer error` | Tokenizer mismatch | Use `--tokenizer` to specify correct tokenizer |\n\n## Best Practices\n\n- Use `--gpu-memory-utilization 0.90` to leave headroom for CUDA kernels.\n- Pin model versions with `--revision` for reproducible deployments.\n- Set `HF_HUB_OFFLINE=1` in production to prevent unexpected downloads.\n- Use AWQ or GPTQ quantization before tensor parallelism — lower VRAM first.\n- Enable `--enable-chunked-prefill` for long-context workloads.\n- Monitor `gpu_cache_usage_perc` — above 95% causes queuing.\n\n## Related Skills\n\n- llm-inference-scaling (`llm-inference-scaling`) - Auto-scaling vLLM deployments\n- gpu-server-management (`gpu-server-management`) - GPU driver setup\n- llm-gateway (`llm-gateway`) - Load balancing across vLLM instances\n- llm-cost-optimization (`llm-cost-optimization`) - Cost management\n- model-serving-kubernetes (`model-serving-kubernetes`) - K8s deployment\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":35,"source":51,"license":14,"category":52,"metadata":53,"date_added":56,"description":7,"source_repo":57,"source_type":39,"compatibility":58,"license_source":59},"https:\u002F\u002Fgithub.com\u002FBagelHole\u002FDevOps-Security-Agent-Skills","devops",{"author":54,"version":55},"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\u002Fvllm-server","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":66,"prompt":76,"zipAvailable":77},[67,70,73],{"tool":68,"command":69},"npx","npx skills add duc-duy-1612\u002FOnline-Delivery-Systems-Web --skill vllm-server",{"tool":71,"command":72},"bunx","bunx skills add duc-duy-1612\u002FOnline-Delivery-Systems-Web --skill vllm-server",{"tool":74,"command":75},"pnpm","pnpm dlx skills add duc-duy-1612\u002FOnline-Delivery-Systems-Web --skill vllm-server","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\u002Fvllm-server\u002FSKILL.md",true,[],[80,98,122,147,172,196],{"skillNo":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":87,"publishedAt":46},"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",[86],{"categoryNo":23,"categoryName":24,"taxonomyType":20,"taxonomyTypeDesc":21},[88,89,90,91,92,93,82,94,44,42,41,40,39,29,95,35,96,34,33,97],"ZeRO","trust","zero trust","architecture","assessment","verification","segmentation","active","scope","implementing",{"skillNo":99,"name":100,"nameZh":6,"description":101,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":102,"categories":103,"keywords":104,"publishedAt":46},"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",[],[105,106,107,108,109,110,111,112,113,114,115,116,117,40,118,39,119,120,35,121],"delegate","zcode","zcode delegate","review","cli","orchestration","tasks","run","Bash","model","node","read","orchestrator","files","desktop app","app","agent-orchestration",{"skillNo":123,"name":124,"nameZh":6,"description":125,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":126,"categories":127,"keywords":128,"publishedAt":46},"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",[],[129,130,131,132,133,134,135,136,137,138,139,140,141,39,29,142,143,144,145,146],"transcript","youtube transcript","youtube","safe","github","codex","research","license","https","text","fetch","tooling","clean","local","output","deepapi","youtube transcripts","blob",{"skillNo":148,"name":149,"nameZh":6,"description":150,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":151,"categories":152,"keywords":162,"publishedAt":46},"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",[153,154,157],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},{"categoryNo":155,"categoryName":156,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000016","图像与多媒体",{"categoryNo":158,"categoryName":159,"taxonomyType":160,"taxonomyTypeDesc":161},"01JBSEEDCAT000000000000030","开发工程师","PROFESSION","适用职业",[129,130,131,163,132,164,133,134,136,165,43,166,167,139,168,39,29,169,170,171],"youtube transcript skills","gemini","search","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":46},"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":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000003","演示文稿",{"categoryNo":155,"categoryName":156,"taxonomyType":20,"taxonomyTypeDesc":21},[131,183,184,109,132,185,186,187,188,136,166,189,190,191,192,29,193,142,194,195],"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":46},"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":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},[204,131,205,206,132,164,165,26,207,43,166,167,139,208,168,39,29,169,209,145],"youtube full","full","google","dlp","credits","automatic"]