[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6DEZC1GPKAVT03JS59Y2Q31":3,"skill-same:01M6DEZC1GPKAVT03JS59Y2Q31":83},{"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":61,"commitSha":62,"skillDirectory":63,"forksCount":64,"repositoryHtmlUrl":65,"repositoryHomepageUrl":6,"ownerHtmlUrl":66,"ownerBlogUrl":6,"ownerDescription":67,"updatedAt":68,"install":69,"resources":82},"01M6DEZC1GPKAVT03JS59Y2Q31","llama-cpp",null,"Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1\u002FM2\u002FM3 Macs, AMD\u002FIntel GPUs, or when CUDA is unavailable. Supports GGUF quantization (1.5-8 bit) for reduced memory and 4-10× speedup vs PyTorch on CPU.","zapabob\u002FSkills","zapabob","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F130830376?v=4","01M6DBMYZAGRTFQDSQYW3W764P","峯岸 亮","01M6DBMZ3GKDSJ01NZ9XTZD7JD",2,"MIT","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Farchive\u002Fskill-md-reviewed\u002Fsha256\u002F43\u002F43ef9b1b96c921ebd1e684a50d26b277cd14caf20f900f6a347c67f81692f8e5\u002FSKILL.md",[],[19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38],"cpp","llama cpp","Llama","apple silicon","bit","apple","cuda","hardware","gguf quantization","embedded","use","llm inference","python","Inference Serving","NVIDIA","edge","deployment","pytorch","research","memory","2026-11-10 10:12:30",false,"---\r\nname: llama-cpp\r\ndescription: Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1\u002FM2\u002FM3 Macs, AMD\u002FIntel GPUs, or when CUDA is unavailable. Supports GGUF quantization (1.5-8 bit) for reduced memory and 4-10× speedup vs PyTorch on CPU.\r\nversion: 1.0.0\r\nauthor: Orchestra Research\r\nlicense: MIT\r\ndependencies: [llama-cpp-python]\r\nmetadata:\r\n  hermes:\r\n    tags: [Inference Serving, Llama.cpp, CPU Inference, Apple Silicon, Edge Deployment, GGUF, Quantization, Non-NVIDIA, AMD GPUs, Intel GPUs, Embedded]\r\n\r\n---\r\n\r\n# llama.cpp\r\n\r\nPure C\u002FC++ LLM inference with minimal dependencies, optimized for CPUs and non-NVIDIA hardware.\r\n\r\n## When to use llama.cpp\r\n\r\n**Use llama.cpp when:**\r\n- Running on CPU-only machines\r\n- Deploying on Apple Silicon (M1\u002FM2\u002FM3\u002FM4)\r\n- Using AMD or Intel GPUs (no CUDA)\r\n- Edge deployment (Raspberry Pi, embedded systems)\r\n- Need simple deployment without Docker\u002FPython\r\n\r\n**Use TensorRT-LLM instead when:**\r\n- Have NVIDIA GPUs (A100\u002FH100)\r\n- Need maximum throughput (100K+ tok\u002Fs)\r\n- Running in datacenter with CUDA\r\n\r\n**Use vLLM instead when:**\r\n- Have NVIDIA GPUs\r\n- Need Python-first API\r\n- Want PagedAttention\r\n\r\n## Quick start\r\n\r\n### Installation\r\n\r\n```bash\r\n# macOS\u002FLinux\r\nbrew install llama.cpp\r\n\r\n# Or build from source\r\ngit clone https:\u002F\u002Fgithub.com\u002Fggerganov\u002Fllama.cpp\r\ncd llama.cpp\r\nmake\r\n\r\n# With Metal (Apple Silicon)\r\nmake LLAMA_METAL=1\r\n\r\n# With CUDA (NVIDIA)\r\nmake LLAMA_CUDA=1\r\n\r\n# With ROCm (AMD)\r\nmake LLAMA_HIP=1\r\n```\r\n\r\n### Download model\r\n\r\n```bash\r\n# Download from HuggingFace (GGUF format)\r\nhuggingface-cli download \\\r\n    TheBloke\u002FLlama-2-7B-Chat-GGUF \\\r\n    llama-2-7b-chat.Q4_K_M.gguf \\\r\n    --local-dir models\u002F\r\n\r\n# Or convert from HuggingFace\r\npython convert_hf_to_gguf.py models\u002Fllama-2-7b-chat\u002F\r\n```\r\n\r\n### Run inference\r\n\r\n```bash\r\n# Simple chat\r\n.\u002Fllama-cli \\\r\n    -m models\u002Fllama-2-7b-chat.Q4_K_M.gguf \\\r\n    -p \"Explain quantum computing\" \\\r\n    -n 256  # Max tokens\r\n\r\n# Interactive chat\r\n.\u002Fllama-cli \\\r\n    -m models\u002Fllama-2-7b-chat.Q4_K_M.gguf \\\r\n    --interactive\r\n```\r\n\r\n### Server mode\r\n\r\n```bash\r\n# Start OpenAI-compatible server\r\n.\u002Fllama-server \\\r\n    -m models\u002Fllama-2-7b-chat.Q4_K_M.gguf \\\r\n    --host 0.0.0.0 \\\r\n    --port 8080 \\\r\n    -ngl 32  # Offload 32 layers to GPU\r\n\r\n# Client request\r\ncurl http:\u002F\u002Flocalhost:8080\u002Fv1\u002Fchat\u002Fcompletions \\\r\n  -H \"Content-Type: application\u002Fjson\" \\\r\n  -d '{\r\n    \"model\": \"llama-2-7b-chat\",\r\n    \"messages\": [{\"role\": \"user\", \"content\": \"Hello!\"}],\r\n    \"temperature\": 0.7,\r\n    \"max_tokens\": 100\r\n  }'\r\n```\r\n\r\n## Quantization formats\r\n\r\n### GGUF format overview\r\n\r\n| Format | Bits | Size (7B) | Speed | Quality | Use Case |\r\n|--------|------|-----------|-------|---------|----------|\r\n| **Q4_K_M** | 4.5 | 4.1 GB | Fast | Good | **Recommended default** |\r\n| Q4_K_S | 4.3 | 3.9 GB | Faster | Lower | Speed critical |\r\n| Q5_K_M | 5.5 | 4.8 GB | Medium | Better | Quality critical |\r\n| Q6_K | 6.5 | 5.5 GB | Slower | Best | Maximum quality |\r\n| Q8_0 | 8.0 | 7.0 GB | Slow | Excellent | Minimal degradation |\r\n| Q2_K | 2.5 | 2.7 GB | Fastest | Poor | Testing only |\r\n\r\n### Choosing quantization\r\n\r\n```bash\r\n# General use (balanced)\r\nQ4_K_M  # 4-bit, medium quality\r\n\r\n# Maximum speed (more degradation)\r\nQ2_K or Q3_K_M\r\n\r\n# Maximum quality (slower)\r\nQ6_K or Q8_0\r\n\r\n# Very large models (70B, 405B)\r\nQ3_K_M or Q4_K_S  # Lower bits to fit in memory\r\n```\r\n\r\n## Hardware acceleration\r\n\r\n### Apple Silicon (Metal)\r\n\r\n```bash\r\n# Build with Metal\r\nmake LLAMA_METAL=1\r\n\r\n# Run with GPU acceleration (automatic)\r\n.\u002Fllama-cli -m model.gguf -ngl 999  # Offload all layers\r\n\r\n# Performance: M3 Max 40-60 tokens\u002Fsec (Llama 2-7B Q4_K_M)\r\n```\r\n\r\n### NVIDIA GPUs (CUDA)\r\n\r\n```bash\r\n# Build with CUDA\r\nmake LLAMA_CUDA=1\r\n\r\n# Offload layers to GPU\r\n.\u002Fllama-cli -m model.gguf -ngl 35  # Offload 35\u002F40 layers\r\n\r\n# Hybrid CPU+GPU for large models\r\n.\u002Fllama-cli -m llama-70b.Q4_K_M.gguf -ngl 20  # GPU: 20 layers, CPU: rest\r\n```\r\n\r\n### AMD GPUs (ROCm)\r\n\r\n```bash\r\n# Build with ROCm\r\nmake LLAMA_HIP=1\r\n\r\n# Run with AMD GPU\r\n.\u002Fllama-cli -m model.gguf -ngl 999\r\n```\r\n\r\n## Common patterns\r\n\r\n### Batch processing\r\n\r\n```bash\r\n# Process multiple prompts from file\r\ncat prompts.txt | .\u002Fllama-cli \\\r\n    -m model.gguf \\\r\n    --batch-size 512 \\\r\n    -n 100\r\n```\r\n\r\n### Constrained generation\r\n\r\n```bash\r\n# JSON output with grammar\r\n.\u002Fllama-cli \\\r\n    -m model.gguf \\\r\n    -p \"Generate a person: \" \\\r\n    --grammar-file grammars\u002Fjson.gbnf\r\n\r\n# Outputs valid JSON only\r\n```\r\n\r\n### Context size\r\n\r\n```bash\r\n# Increase context (default 512)\r\n.\u002Fllama-cli \\\r\n    -m model.gguf \\\r\n    -c 4096  # 4K context window\r\n\r\n# Very long context (if model supports)\r\n.\u002Fllama-cli -m model.gguf -c 32768  # 32K context\r\n```\r\n\r\n## Performance benchmarks\r\n\r\n### CPU performance (Llama 2-7B Q4_K_M)\r\n\r\n| CPU | Threads | Speed | Cost |\r\n|-----|---------|-------|------|\r\n| Apple M3 Max | 16 | 50 tok\u002Fs | $0 (local) |\r\n| AMD Ryzen 9 7950X | 32 | 35 tok\u002Fs | $0.50\u002Fhour |\r\n| Intel i9-13900K | 32 | 30 tok\u002Fs | $0.40\u002Fhour |\r\n| AWS c7i.16xlarge | 64 | 40 tok\u002Fs | $2.88\u002Fhour |\r\n\r\n### GPU acceleration (Llama 2-7B Q4_K_M)\r\n\r\n| GPU | Speed | vs CPU | Cost |\r\n|-----|-------|--------|------|\r\n| NVIDIA RTX 4090 | 120 tok\u002Fs | 3-4× | $0 (local) |\r\n| NVIDIA A10 | 80 tok\u002Fs | 2-3× | $1.00\u002Fhour |\r\n| AMD MI250 | 70 tok\u002Fs | 2× | $2.00\u002Fhour |\r\n| Apple M3 Max (Metal) | 50 tok\u002Fs | ~Same | $0 (local) |\r\n\r\n## Supported models\r\n\r\n**LLaMA family**:\r\n- Llama 2 (7B, 13B, 70B)\r\n- Llama 3 (8B, 70B, 405B)\r\n- Code Llama\r\n\r\n**Mistral family**:\r\n- Mistral 7B\r\n- Mixtral 8x7B, 8x22B\r\n\r\n**Other**:\r\n- Falcon, BLOOM, GPT-J\r\n- Phi-3, Gemma, Qwen\r\n- LLaVA (vision), Whisper (audio)\r\n\r\n**Find models**: https:\u002F\u002Fhuggingface.co\u002Fmodels?library=gguf\r\n\r\n## References\r\n\r\n- **[Quantization Guide](references\u002Fquantization.md)** - GGUF formats, conversion, quality comparison\r\n- **[Server Deployment](references\u002Fserver.md)** - API endpoints, Docker, monitoring\r\n- **[Optimization](references\u002Foptimization.md)** - Performance tuning, hybrid CPU+GPU\r\n\r\n## Resources\r\n\r\n- **GitHub**: https:\u002F\u002Fgithub.com\u002Fggerganov\u002Fllama.cpp\r\n- **Models**: https:\u002F\u002Fhuggingface.co\u002Fmodels?library=gguf\r\n- **Discord**: https:\u002F\u002Fdiscord.gg\u002Fllama-cpp\r\n","# llama.cpp\r\n\r\nPure C\u002FC++ LLM inference with minimal dependencies, optimized for CPUs and non-NVIDIA hardware.\r\n\r\n## When to use llama.cpp\r\n\r\n**Use llama.cpp when:**\r\n- Running on CPU-only machines\r\n- Deploying on Apple Silicon (M1\u002FM2\u002FM3\u002FM4)\r\n- Using AMD or Intel GPUs (no CUDA)\r\n- Edge deployment (Raspberry Pi, embedded systems)\r\n- Need simple deployment without Docker\u002FPython\r\n\r\n**Use TensorRT-LLM instead when:**\r\n- Have NVIDIA GPUs (A100\u002FH100)\r\n- Need maximum throughput (100K+ tok\u002Fs)\r\n- Running in datacenter with CUDA\r\n\r\n**Use vLLM instead when:**\r\n- Have NVIDIA GPUs\r\n- Need Python-first API\r\n- Want PagedAttention\r\n\r\n## Quick start\r\n\r\n### Installation\r\n\r\n```bash\r\n# macOS\u002FLinux\r\nbrew install llama.cpp\r\n\r\n# Or build from source\r\ngit clone https:\u002F\u002Fgithub.com\u002Fggerganov\u002Fllama.cpp\r\ncd llama.cpp\r\nmake\r\n\r\n# With Metal (Apple Silicon)\r\nmake LLAMA_METAL=1\r\n\r\n# With CUDA (NVIDIA)\r\nmake LLAMA_CUDA=1\r\n\r\n# With ROCm (AMD)\r\nmake LLAMA_HIP=1\r\n```\r\n\r\n### Download model\r\n\r\n```bash\r\n# Download from HuggingFace (GGUF format)\r\nhuggingface-cli download \\\r\n    TheBloke\u002FLlama-2-7B-Chat-GGUF \\\r\n    llama-2-7b-chat.Q4_K_M.gguf \\\r\n    --local-dir models\u002F\r\n\r\n# Or convert from HuggingFace\r\npython convert_hf_to_gguf.py models\u002Fllama-2-7b-chat\u002F\r\n```\r\n\r\n### Run inference\r\n\r\n```bash\r\n# Simple chat\r\n.\u002Fllama-cli \\\r\n    -m models\u002Fllama-2-7b-chat.Q4_K_M.gguf \\\r\n    -p \"Explain quantum computing\" \\\r\n    -n 256  # Max tokens\r\n\r\n# Interactive chat\r\n.\u002Fllama-cli \\\r\n    -m models\u002Fllama-2-7b-chat.Q4_K_M.gguf \\\r\n    --interactive\r\n```\r\n\r\n### Server mode\r\n\r\n```bash\r\n# Start OpenAI-compatible server\r\n.\u002Fllama-server \\\r\n    -m models\u002Fllama-2-7b-chat.Q4_K_M.gguf \\\r\n    --host 0.0.0.0 \\\r\n    --port 8080 \\\r\n    -ngl 32  # Offload 32 layers to GPU\r\n\r\n# Client request\r\ncurl http:\u002F\u002Flocalhost:8080\u002Fv1\u002Fchat\u002Fcompletions \\\r\n  -H \"Content-Type: application\u002Fjson\" \\\r\n  -d '{\r\n    \"model\": \"llama-2-7b-chat\",\r\n    \"messages\": [{\"role\": \"user\", \"content\": \"Hello!\"}],\r\n    \"temperature\": 0.7,\r\n    \"max_tokens\": 100\r\n  }'\r\n```\r\n\r\n## Quantization formats\r\n\r\n### GGUF format overview\r\n\r\n| Format | Bits | Size (7B) | Speed | Quality | Use Case |\r\n|--------|------|-----------|-------|---------|----------|\r\n| **Q4_K_M** | 4.5 | 4.1 GB | Fast | Good | **Recommended default** |\r\n| Q4_K_S | 4.3 | 3.9 GB | Faster | Lower | Speed critical |\r\n| Q5_K_M | 5.5 | 4.8 GB | Medium | Better | Quality critical |\r\n| Q6_K | 6.5 | 5.5 GB | Slower | Best | Maximum quality |\r\n| Q8_0 | 8.0 | 7.0 GB | Slow | Excellent | Minimal degradation |\r\n| Q2_K | 2.5 | 2.7 GB | Fastest | Poor | Testing only |\r\n\r\n### Choosing quantization\r\n\r\n```bash\r\n# General use (balanced)\r\nQ4_K_M  # 4-bit, medium quality\r\n\r\n# Maximum speed (more degradation)\r\nQ2_K or Q3_K_M\r\n\r\n# Maximum quality (slower)\r\nQ6_K or Q8_0\r\n\r\n# Very large models (70B, 405B)\r\nQ3_K_M or Q4_K_S  # Lower bits to fit in memory\r\n```\r\n\r\n## Hardware acceleration\r\n\r\n### Apple Silicon (Metal)\r\n\r\n```bash\r\n# Build with Metal\r\nmake LLAMA_METAL=1\r\n\r\n# Run with GPU acceleration (automatic)\r\n.\u002Fllama-cli -m model.gguf -ngl 999  # Offload all layers\r\n\r\n# Performance: M3 Max 40-60 tokens\u002Fsec (Llama 2-7B Q4_K_M)\r\n```\r\n\r\n### NVIDIA GPUs (CUDA)\r\n\r\n```bash\r\n# Build with CUDA\r\nmake LLAMA_CUDA=1\r\n\r\n# Offload layers to GPU\r\n.\u002Fllama-cli -m model.gguf -ngl 35  # Offload 35\u002F40 layers\r\n\r\n# Hybrid CPU+GPU for large models\r\n.\u002Fllama-cli -m llama-70b.Q4_K_M.gguf -ngl 20  # GPU: 20 layers, CPU: rest\r\n```\r\n\r\n### AMD GPUs (ROCm)\r\n\r\n```bash\r\n# Build with ROCm\r\nmake LLAMA_HIP=1\r\n\r\n# Run with AMD GPU\r\n.\u002Fllama-cli -m model.gguf -ngl 999\r\n```\r\n\r\n## Common patterns\r\n\r\n### Batch processing\r\n\r\n```bash\r\n# Process multiple prompts from file\r\ncat prompts.txt | .\u002Fllama-cli \\\r\n    -m model.gguf \\\r\n    --batch-size 512 \\\r\n    -n 100\r\n```\r\n\r\n### Constrained generation\r\n\r\n```bash\r\n# JSON output with grammar\r\n.\u002Fllama-cli \\\r\n    -m model.gguf \\\r\n    -p \"Generate a person: \" \\\r\n    --grammar-file grammars\u002Fjson.gbnf\r\n\r\n# Outputs valid JSON only\r\n```\r\n\r\n### Context size\r\n\r\n```bash\r\n# Increase context (default 512)\r\n.\u002Fllama-cli \\\r\n    -m model.gguf \\\r\n    -c 4096  # 4K context window\r\n\r\n# Very long context (if model supports)\r\n.\u002Fllama-cli -m model.gguf -c 32768  # 32K context\r\n```\r\n\r\n## Performance benchmarks\r\n\r\n### CPU performance (Llama 2-7B Q4_K_M)\r\n\r\n| CPU | Threads | Speed | Cost |\r\n|-----|---------|-------|------|\r\n| Apple M3 Max | 16 | 50 tok\u002Fs | $0 (local) |\r\n| AMD Ryzen 9 7950X | 32 | 35 tok\u002Fs | $0.50\u002Fhour |\r\n| Intel i9-13900K | 32 | 30 tok\u002Fs | $0.40\u002Fhour |\r\n| AWS c7i.16xlarge | 64 | 40 tok\u002Fs | $2.88\u002Fhour |\r\n\r\n### GPU acceleration (Llama 2-7B Q4_K_M)\r\n\r\n| GPU | Speed | vs CPU | Cost |\r\n|-----|-------|--------|------|\r\n| NVIDIA RTX 4090 | 120 tok\u002Fs | 3-4× | $0 (local) |\r\n| NVIDIA A10 | 80 tok\u002Fs | 2-3× | $1.00\u002Fhour |\r\n| AMD MI250 | 70 tok\u002Fs | 2× | $2.00\u002Fhour |\r\n| Apple M3 Max (Metal) | 50 tok\u002Fs | ~Same | $0 (local) |\r\n\r\n## Supported models\r\n\r\n**LLaMA family**:\r\n- Llama 2 (7B, 13B, 70B)\r\n- Llama 3 (8B, 70B, 405B)\r\n- Code Llama\r\n\r\n**Mistral family**:\r\n- Mistral 7B\r\n- Mixtral 8x7B, 8x22B\r\n\r\n**Other**:\r\n- Falcon, BLOOM, GPT-J\r\n- Phi-3, Gemma, Qwen\r\n- LLaVA (vision), Whisper (audio)\r\n\r\n**Find models**: https:\u002F\u002Fhuggingface.co\u002Fmodels?library=gguf\r\n\r\n## References\r\n\r\n- **[Quantization Guide](references\u002Fquantization.md)** - GGUF formats, conversion, quality comparison\r\n- **[Server Deployment](references\u002Fserver.md)** - API endpoints, Docker, monitoring\r\n- **[Optimization](references\u002Foptimization.md)** - Performance tuning, hybrid CPU+GPU\r\n\r\n## Resources\r\n\r\n- **GitHub**: https:\u002F\u002Fgithub.com\u002Fggerganov\u002Fllama.cpp\r\n- **Models**: https:\u002F\u002Fhuggingface.co\u002Fmodels?library=gguf\r\n- **Discord**: https:\u002F\u002Fdiscord.gg\u002Fllama-cpp\r\n",{"name":5,"author":44,"license":15,"version":45,"metadata":46,"description":7,"dependencies":59},"Orchestra Research","1.0.0",{"hermes":47},{"tags":48},[32,49,50,51,52,53,54,55,56,57,58],"Llama.cpp","CPU Inference","Apple Silicon","Edge Deployment","GGUF","Quantization","Non-NVIDIA","AMD GPUs","Intel GPUs","Embedded",[60],"llama-cpp-python","main","0d1434f89f4bc6b07e9d272accd8c1013768820f","archive\u002Fskill-md-reviewed\u002Fsha256\u002F43\u002F43ef9b1b96c921ebd1e684a50d26b277cd14caf20f900f6a347c67f81692f8e5",0,"https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills","https:\u002F\u002Fgithub.com\u002Fzapabob","The Open University of JAPAN　2019-now\r\n\r\nTokyo University of Science　2007-2010","2026-10-26 16:29:01",{"source":8,"skillName":5,"commands":70,"prompt":80,"zipAvailable":81},[71,74,77],{"tool":72,"command":73},"npx","npx skills add zapabob\u002FSkills --skill llama-cpp",{"tool":75,"command":76},"bunx","bunx skills add zapabob\u002FSkills --skill llama-cpp",{"tool":78,"command":79},"pnpm","pnpm dlx skills add zapabob\u002FSkills --skill llama-cpp","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Farchive\u002Fskill-md-reviewed\u002Fsha256\u002F43\u002F43ef9b1b96c921ebd1e684a50d26b277cd14caf20f900f6a347c67f81692f8e5\u002FSKILL.md",true,[],[84,122,157,183,209,226],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":103,"publishedAt":39},"01M6GJTE1GESGGF9K028Y7753E","xurl","X\u002FTwitter via xurl CLI: post, search, DM, media, v2 API.","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Fregistry\u002Fskills\u002Fxurl\u002Fvariants\u002Fhermes-agent\u002FSKILL.md",[90,95,98],{"categoryNo":91,"categoryName":92,"taxonomyType":93,"taxonomyTypeDesc":94},"01JBSEEDCAT000000000000010","后端与接口","SCENARIO","使用场景",{"categoryNo":96,"categoryName":97,"taxonomyType":93,"taxonomyTypeDesc":94},"01JBSEEDCAT000000000000023","Skill与智能体开发",{"categoryNo":99,"categoryName":100,"taxonomyType":101,"taxonomyTypeDesc":102},"01JBSEEDCAT000000000000030","开发工程师","PROFESSION","适用职业",[86,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,61,119,120,121],"openclaw","macos","cli","github","linux","windows","official","search","https","social-media","api","Media","hermes","hermes agent","twitter","social","via","post",{"skillNo":123,"name":124,"nameZh":6,"description":125,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":126,"categories":127,"keywords":139,"publishedAt":39},"01M6GJT8S1PXQ8WPXT1HSNBWH2","writing-plans","Write implementation plans: bite-sized tasks, paths, code.","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Fregistry\u002Fskills\u002Fwriting-plans\u002Fvariants\u002Fhermes-agent\u002FSKILL.md",[128,131,134,135,136],{"categoryNo":129,"categoryName":130,"taxonomyType":93,"taxonomyTypeDesc":94},"01JBSEEDCAT000000000000015","视觉设计",{"categoryNo":132,"categoryName":133,"taxonomyType":93,"taxonomyTypeDesc":94},"01JBSEEDCAT000000000000018","写作与文案",{"categoryNo":96,"categoryName":97,"taxonomyType":93,"taxonomyTypeDesc":94},{"categoryNo":99,"categoryName":100,"taxonomyType":101,"taxonomyTypeDesc":102},{"categoryNo":137,"categoryName":138,"taxonomyType":101,"taxonomyTypeDesc":102},"01JBSEEDCAT000000000000036","内容创作者",[140,141,142,143,105,144,108,145,109,146,147,148,149,150,151,152,153,154,155,156],"writing","writing plans","plans","design","review","workflow","documentation","tasks","development","code","planning","code-review","subagent","write","superpowers","paths","test-driven-development",{"skillNo":158,"name":159,"nameZh":6,"description":160,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":161,"categories":162,"keywords":166,"publishedAt":39},"01M6GJST0H5X7WHBJ3BZK0DVKH","whisper","OpenAI's general-purpose speech recognition model. Supports 99 languages, transcription, translation to English, and language identification. Six model sizes from tiny (39M params) to large (1550M params). Use for speech-to-text, podcast transcription, or multilingual audio processing. Best for robust, multilingual ASR.","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Fregistry\u002Fskills\u002Fwhisper\u002Fvariants\u002Fhermes-agent-optional\u002FSKILL.md",[163],{"categoryNo":164,"categoryName":165,"taxonomyType":93,"taxonomyTypeDesc":94},"01JBSEEDCAT000000000000016","图像与多媒体",[159,105,108,167,168,169,170,171,172,173,174,175,29,176,177,178,179,180,181,182],"openai","speech","multimodal","Transformers","ASR","Audio","text","model","speech recognition","general","multilingual","translation","english","identification","processing","Audio Processing",{"skillNo":184,"name":185,"nameZh":6,"description":186,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":187,"categories":188,"keywords":192,"publishedAt":39},"01M6GJSMRQ8EETZTYMG72P70EA","weights-and-biases","W&B: log ML experiments, sweeps, model registry, dashboards.","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Fregistry\u002Fskills\u002Fweights-and-biases\u002Fvariants\u002Fhermes-agent\u002FSKILL.md",[189],{"categoryNo":190,"categoryName":191,"taxonomyType":93,"taxonomyTypeDesc":94},"01JBSEEDCAT000000000000013","分析与可视化",[193,105,194,195,109,36,196,197,198,199,200,174,201,202,203,204,205,206,207,208],"weights and biases","tensorflow","experiments","visualization","registry","real-time","tuning","dashboards","experiment","log","time","Experiment Tracking","Model Registry","Hyperparameter Tuning","WandB","Real-Time Visualization",{"skillNo":210,"name":211,"nameZh":6,"description":212,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":213,"categories":214,"keywords":216,"publishedAt":39},"01M6GJS19KDPRY112JTR326S06","watchers","Poll RSS, JSON APIs, and GitHub with watermark dedup.","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Fregistry\u002Fskills\u002Fwatchers\u002Fvariants\u002Fhermes-agent-optional\u002FSKILL.md",[215],{"categoryNo":96,"categoryName":97,"taxonomyType":93,"taxonomyTypeDesc":94},[211,105,217,107,108,218,219,220,221,222,223,116,117,224,225],"terminal","automation","http","monitoring","cron","json","rss","devops","with",{"skillNo":227,"name":228,"nameZh":6,"description":229,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":230,"categories":231,"keywords":234,"publishedAt":39},"01M6GJRHJD2A6Q830SWF9W48BC","serving-llms-vllm","vLLM: high-throughput LLM serving, OpenAI API, quantization.","https:\u002F\u002Fgithub.com\u002Fzapabob\u002FSkills\u002Fblob\u002Fmain\u002Fregistry\u002Fskills\u002Fvllm\u002Fvariants\u002Fhermes-agent\u002FSKILL.md",[232,233],{"categoryNo":91,"categoryName":92,"taxonomyType":93,"taxonomyTypeDesc":94},{"categoryNo":99,"categoryName":100,"taxonomyType":101,"taxonomyTypeDesc":102},[235,236,237,108,37,167,238,239,32,170,240,114,241,242,243,244,54,245,246,247],"llms","vLLM","serving llms vllm","openai api","tensor","Production","llm serving","High Throughput","PagedAttention","Continuous Batching","Tensor Parallelism","high-throughput","throughput"]