[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M74FQTP9C6J4EFHTS2FPZ5D4":3,"skill-same:01M74FQTP9C6J4EFHTS2FPZ5D4":91},{"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":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":45,"commitSha":46,"skillDirectory":47,"forksCount":13,"repositoryHtmlUrl":48,"repositoryHomepageUrl":6,"ownerHtmlUrl":49,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":50,"install":51,"resources":64},"01M74FQTP9C6J4EFHTS2FPZ5D4","llama-cpp",null,"llama.cpp local GGUF inference + HF Hub model discovery.","lzy18001500226\u002FMoSim","lzy18001500226","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F152623587?v=4","01M71ZMEX36NK05B642131DEEK","01M71ZMEXSMX72KPVNJ572G0CM",0,"MIT","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002FReferences\u002FAgent\u002FPlatforms\u002Fhermes-agent\u002Fskills\u002Fmlops\u002Finference\u002Fllama-cpp\u002FSKILL.md",[],[],"2026-11-11 15:11:14",false,"---\nname: llama-cpp\ndescription: llama.cpp local GGUF inference + HF Hub model discovery.\nversion: 2.1.2\nauthor: Orchestra Research\nlicense: MIT\ndependencies: [llama-cpp-python>=0.2.0]\nplatforms: [linux, macos, windows]\nmetadata:\n  hermes:\n    tags: [llama.cpp, GGUF, Quantization, Hugging Face Hub, CPU Inference, Apple Silicon, Edge Deployment, AMD GPUs, Intel GPUs, NVIDIA, URL-first]\n---\n\n# llama.cpp + GGUF\n\nUse this skill for local GGUF inference, quant selection, or Hugging Face repo discovery for llama.cpp.\n\n## When to use\n\n- Run local models on CPU, Apple Silicon, CUDA, ROCm, or Intel GPUs\n- Find the right GGUF for a specific Hugging Face repo\n- Build a `llama-server` or `llama-cli` command from the Hub\n- Search the Hub for models that already support llama.cpp\n- Enumerate available `.gguf` files and sizes for a repo\n- Decide between Q4\u002FQ5\u002FQ6\u002FIQ variants for the user's RAM or VRAM\n\n## Model Discovery workflow\n\nPrefer URL workflows before asking for `hf`, Python, or custom scripts.\n\n1. Search for candidate repos on the Hub:\n   - Base: `https:\u002F\u002Fhuggingface.co\u002Fmodels?apps=llama.cpp&sort=trending`\n   - Add `search=\u003Cterm>` for a model family\n   - Add `num_parameters=min:0,max:24B` or similar when the user has size constraints\n2. Open the repo with the llama.cpp local-app view:\n   - `https:\u002F\u002Fhuggingface.co\u002F\u003Crepo>?local-app=llama.cpp`\n3. Treat the local-app snippet as the source of truth when it is visible:\n   - copy the exact `llama-server` or `llama-cli` command\n   - report the recommended quant exactly as HF shows it\n4. Read the same `?local-app=llama.cpp` URL as page text or HTML and extract the section under `Hardware compatibility`:\n   - prefer its exact quant labels and sizes over generic tables\n   - keep repo-specific labels such as `UD-Q4_K_M` or `IQ4_NL_XL`\n   - if that section is not visible in the fetched page source, say so and fall back to the tree API plus generic quant guidance\n5. Query the tree API to confirm what actually exists:\n   - `https:\u002F\u002Fhuggingface.co\u002Fapi\u002Fmodels\u002F\u003Crepo>\u002Ftree\u002Fmain?recursive=true`\n   - keep entries where `type` is `file` and `path` ends with `.gguf`\n   - use `path` and `size` as the source of truth for filenames and byte sizes\n   - separate quantized checkpoints from `mmproj-*.gguf` projector files and `BF16\u002F` shard files\n   - use `https:\u002F\u002Fhuggingface.co\u002F\u003Crepo>\u002Ftree\u002Fmain` only as a human fallback\n6. If the local-app snippet is not text-visible, reconstruct the command from the repo plus the chosen quant:\n   - shorthand quant selection: `llama-server -hf \u003Crepo>:\u003CQUANT>`\n   - exact-file fallback: `llama-server --hf-repo \u003Crepo> --hf-file \u003Cfilename.gguf>`\n7. Only suggest conversion from Transformers weights if the repo does not already expose GGUF files.\n\n## Quick start\n\n### Install llama.cpp\n\n```bash\n# macOS \u002F Linux (simplest)\nbrew install llama.cpp\n```\n\n```bash\nwinget install llama.cpp\n```\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fggml-org\u002Fllama.cpp\ncd llama.cpp\ncmake -B build\ncmake --build build --config Release\n```\n\n### Run directly from the Hugging Face Hub\n\n```bash\nllama-cli -hf bartowski\u002FLlama-3.2-3B-Instruct-GGUF:Q8_0\n```\n\n```bash\nllama-server -hf bartowski\u002FLlama-3.2-3B-Instruct-GGUF:Q8_0\n```\n\n### Run an exact GGUF file from the Hub\n\nUse this when the tree API shows custom file naming or the exact HF snippet is missing.\n\n```bash\nllama-server \\\n    --hf-repo microsoft\u002FPhi-3-mini-4k-instruct-gguf \\\n    --hf-file Phi-3-mini-4k-instruct-q4.gguf \\\n    -c 4096\n```\n\n### OpenAI-compatible server check\n\n```bash\ncurl http:\u002F\u002Flocalhost:8080\u002Fv1\u002Fchat\u002Fcompletions \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\n    \"messages\": [\n      {\"role\": \"user\", \"content\": \"Write a limerick about Python exceptions\"}\n    ]\n  }'\n```\n\n## Python bindings (llama-cpp-python)\n\n`pip install llama-cpp-python` (CUDA: `CMAKE_ARGS=\"-DGGML_CUDA=on\" pip install llama-cpp-python --force-reinstall --no-cache-dir`; Metal: `CMAKE_ARGS=\"-DGGML_METAL=on\" ...`).\n\n### Basic generation\n\n```python\nfrom llama_cpp import Llama\n\nllm = Llama(\n    model_path=\".\u002Fmodel-q4_k_m.gguf\",\n    n_ctx=4096,\n    n_gpu_layers=35,     # 0 for CPU, 99 to offload everything\n    n_threads=8,\n)\n\nout = llm(\"What is machine learning?\", max_tokens=256, temperature=0.7)\nprint(out[\"choices\"][0][\"text\"])\n```\n\n### Chat + streaming\n\n```python\nllm = Llama(\n    model_path=\".\u002Fmodel-q4_k_m.gguf\",\n    n_ctx=4096,\n    n_gpu_layers=35,\n    chat_format=\"llama-3\",   # or \"chatml\", \"mistral\", etc.\n)\n\nresp = llm.create_chat_completion(\n    messages=[\n        {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n        {\"role\": \"user\", \"content\": \"What is Python?\"},\n    ],\n    max_tokens=256,\n)\nprint(resp[\"choices\"][0][\"message\"][\"content\"])\n\n# Streaming\nfor chunk in llm(\"Explain quantum computing:\", max_tokens=256, stream=True):\n    print(chunk[\"choices\"][0][\"text\"], end=\"\", flush=True)\n```\n\n### Embeddings\n\n```python\nllm = Llama(model_path=\".\u002Fmodel-q4_k_m.gguf\", embedding=True, n_gpu_layers=35)\nvec = llm.embed(\"This is a test sentence.\")\nprint(f\"Embedding dimension: {len(vec)}\")\n```\n\nYou can also load a GGUF straight from the Hub:\n\n```python\nllm = Llama.from_pretrained(\n    repo_id=\"bartowski\u002FLlama-3.2-3B-Instruct-GGUF\",\n    filename=\"*Q4_K_M.gguf\",\n    n_gpu_layers=35,\n)\n```\n\n## Choosing a quant\n\nUse the Hub page first, generic heuristics second.\n\n- Prefer the exact quant that HF marks as compatible for the user's hardware profile.\n- For general chat, start with `Q4_K_M`.\n- For code or technical work, prefer `Q5_K_M` or `Q6_K` if memory allows.\n- For very tight RAM budgets, consider `Q3_K_M`, `IQ` variants, or `Q2` variants only if the user explicitly prioritizes fit over quality.\n- For multimodal repos, mention `mmproj-*.gguf` separately. The projector is not the main model file.\n- Do not normalize repo-native labels. If the page says `UD-Q4_K_M`, report `UD-Q4_K_M`.\n\n## Extracting available GGUFs from a repo\n\nWhen the user asks what GGUFs exist, return:\n\n- filename\n- file size\n- quant label\n- whether it is a main model or an auxiliary projector\n\nIgnore unless requested:\n\n- README\n- BF16 shard files\n- imatrix blobs or calibration artifacts\n\nUse the tree API for this step:\n\n- `https:\u002F\u002Fhuggingface.co\u002Fapi\u002Fmodels\u002F\u003Crepo>\u002Ftree\u002Fmain?recursive=true`\n\nFor a repo like `unsloth\u002FQwen3.6-35B-A3B-GGUF`, the local-app page can show quant chips such as `UD-Q4_K_M`, `UD-Q5_K_M`, `UD-Q6_K`, and `Q8_0`, while the tree API exposes exact file paths such as `Qwen3.6-35B-A3B-UD-Q4_K_M.gguf` and `Qwen3.6-35B-A3B-Q8_0.gguf` with byte sizes. Use the tree API to turn a quant label into an exact filename.\n\n## Search patterns\n\nUse these URL shapes directly:\n\n```text\nhttps:\u002F\u002Fhuggingface.co\u002Fmodels?apps=llama.cpp&sort=trending\nhttps:\u002F\u002Fhuggingface.co\u002Fmodels?search=\u003Cterm>&apps=llama.cpp&sort=trending\nhttps:\u002F\u002Fhuggingface.co\u002Fmodels?search=\u003Cterm>&apps=llama.cpp&num_parameters=min:0,max:24B&sort=trending\nhttps:\u002F\u002Fhuggingface.co\u002F\u003Crepo>?local-app=llama.cpp\nhttps:\u002F\u002Fhuggingface.co\u002Fapi\u002Fmodels\u002F\u003Crepo>\u002Ftree\u002Fmain?recursive=true\nhttps:\u002F\u002Fhuggingface.co\u002F\u003Crepo>\u002Ftree\u002Fmain\n```\n\n## Output format\n\nWhen answering discovery requests, prefer a compact structured result like:\n\n```text\nRepo: \u003Crepo>\nRecommended quant from HF: \u003Clabel> (\u003Csize>)\nllama-server: \u003Ccommand>\nOther GGUFs:\n- \u003Cfilename> - \u003Csize>\n- \u003Cfilename> - \u003Csize>\nSource URLs:\n- \u003Clocal-app URL>\n- \u003Ctree API URL>\n```\n\n## References\n\n- **[hub-discovery.md](references\u002Fhub-discovery.md)** - URL-only Hugging Face workflows, search patterns, GGUF extraction, and command reconstruction\n- **[advanced-usage.md](references\u002Fadvanced-usage.md)** — speculative decoding, batched inference, grammar-constrained generation, LoRA, multi-GPU, custom builds, benchmark scripts\n- **[quantization.md](references\u002Fquantization.md)** — quant quality tradeoffs, when to use Q4\u002FQ5\u002FQ6\u002FIQ, model size scaling, imatrix\n- **[server.md](references\u002Fserver.md)** — direct-from-Hub server launch, OpenAI API endpoints, Docker deployment, NGINX load balancing, monitoring\n- **[optimization.md](references\u002Foptimization.md)** — CPU threading, BLAS, GPU offload heuristics, batch tuning, benchmarks\n- **[troubleshooting.md](references\u002Ftroubleshooting.md)** — install\u002Fconvert\u002Fquantize\u002Finference\u002Fserver issues, Apple Silicon, debugging\n\n## Resources\n\n- **GitHub**: https:\u002F\u002Fgithub.com\u002Fggml-org\u002Fllama.cpp\n- **Hugging Face GGUF + llama.cpp docs**: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fgguf-llamacpp\n- **Hugging Face Local Apps docs**: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fmain\u002Flocal-apps\n- **Hugging Face Local Agents docs**: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fagents-local\n- **Example local-app page**: https:\u002F\u002Fhuggingface.co\u002Funsloth\u002FQwen3.6-35B-A3B-GGUF?local-app=llama.cpp\n- **Example tree API**: https:\u002F\u002Fhuggingface.co\u002Fapi\u002Fmodels\u002Funsloth\u002FQwen3.6-35B-A3B-GGUF\u002Ftree\u002Fmain?recursive=true\n- **Example llama.cpp search**: https:\u002F\u002Fhuggingface.co\u002Fmodels?num_parameters=min:0,max:24B&apps=llama.cpp&sort=trending\n- **License**: MIT\n","# llama.cpp + GGUF\n\nUse this skill for local GGUF inference, quant selection, or Hugging Face repo discovery for llama.cpp.\n\n## When to use\n\n- Run local models on CPU, Apple Silicon, CUDA, ROCm, or Intel GPUs\n- Find the right GGUF for a specific Hugging Face repo\n- Build a `llama-server` or `llama-cli` command from the Hub\n- Search the Hub for models that already support llama.cpp\n- Enumerate available `.gguf` files and sizes for a repo\n- Decide between Q4\u002FQ5\u002FQ6\u002FIQ variants for the user's RAM or VRAM\n\n## Model Discovery workflow\n\nPrefer URL workflows before asking for `hf`, Python, or custom scripts.\n\n1. Search for candidate repos on the Hub:\n   - Base: `https:\u002F\u002Fhuggingface.co\u002Fmodels?apps=llama.cpp&sort=trending`\n   - Add `search=\u003Cterm>` for a model family\n   - Add `num_parameters=min:0,max:24B` or similar when the user has size constraints\n2. Open the repo with the llama.cpp local-app view:\n   - `https:\u002F\u002Fhuggingface.co\u002F\u003Crepo>?local-app=llama.cpp`\n3. Treat the local-app snippet as the source of truth when it is visible:\n   - copy the exact `llama-server` or `llama-cli` command\n   - report the recommended quant exactly as HF shows it\n4. Read the same `?local-app=llama.cpp` URL as page text or HTML and extract the section under `Hardware compatibility`:\n   - prefer its exact quant labels and sizes over generic tables\n   - keep repo-specific labels such as `UD-Q4_K_M` or `IQ4_NL_XL`\n   - if that section is not visible in the fetched page source, say so and fall back to the tree API plus generic quant guidance\n5. Query the tree API to confirm what actually exists:\n   - `https:\u002F\u002Fhuggingface.co\u002Fapi\u002Fmodels\u002F\u003Crepo>\u002Ftree\u002Fmain?recursive=true`\n   - keep entries where `type` is `file` and `path` ends with `.gguf`\n   - use `path` and `size` as the source of truth for filenames and byte sizes\n   - separate quantized checkpoints from `mmproj-*.gguf` projector files and `BF16\u002F` shard files\n   - use `https:\u002F\u002Fhuggingface.co\u002F\u003Crepo>\u002Ftree\u002Fmain` only as a human fallback\n6. If the local-app snippet is not text-visible, reconstruct the command from the repo plus the chosen quant:\n   - shorthand quant selection: `llama-server -hf \u003Crepo>:\u003CQUANT>`\n   - exact-file fallback: `llama-server --hf-repo \u003Crepo> --hf-file \u003Cfilename.gguf>`\n7. Only suggest conversion from Transformers weights if the repo does not already expose GGUF files.\n\n## Quick start\n\n### Install llama.cpp\n\n```bash\n# macOS \u002F Linux (simplest)\nbrew install llama.cpp\n```\n\n```bash\nwinget install llama.cpp\n```\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fggml-org\u002Fllama.cpp\ncd llama.cpp\ncmake -B build\ncmake --build build --config Release\n```\n\n### Run directly from the Hugging Face Hub\n\n```bash\nllama-cli -hf bartowski\u002FLlama-3.2-3B-Instruct-GGUF:Q8_0\n```\n\n```bash\nllama-server -hf bartowski\u002FLlama-3.2-3B-Instruct-GGUF:Q8_0\n```\n\n### Run an exact GGUF file from the Hub\n\nUse this when the tree API shows custom file naming or the exact HF snippet is missing.\n\n```bash\nllama-server \\\n    --hf-repo microsoft\u002FPhi-3-mini-4k-instruct-gguf \\\n    --hf-file Phi-3-mini-4k-instruct-q4.gguf \\\n    -c 4096\n```\n\n### OpenAI-compatible server check\n\n```bash\ncurl http:\u002F\u002Flocalhost:8080\u002Fv1\u002Fchat\u002Fcompletions \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\n    \"messages\": [\n      {\"role\": \"user\", \"content\": \"Write a limerick about Python exceptions\"}\n    ]\n  }'\n```\n\n## Python bindings (llama-cpp-python)\n\n`pip install llama-cpp-python` (CUDA: `CMAKE_ARGS=\"-DGGML_CUDA=on\" pip install llama-cpp-python --force-reinstall --no-cache-dir`; Metal: `CMAKE_ARGS=\"-DGGML_METAL=on\" ...`).\n\n### Basic generation\n\n```python\nfrom llama_cpp import Llama\n\nllm = Llama(\n    model_path=\".\u002Fmodel-q4_k_m.gguf\",\n    n_ctx=4096,\n    n_gpu_layers=35,     # 0 for CPU, 99 to offload everything\n    n_threads=8,\n)\n\nout = llm(\"What is machine learning?\", max_tokens=256, temperature=0.7)\nprint(out[\"choices\"][0][\"text\"])\n```\n\n### Chat + streaming\n\n```python\nllm = Llama(\n    model_path=\".\u002Fmodel-q4_k_m.gguf\",\n    n_ctx=4096,\n    n_gpu_layers=35,\n    chat_format=\"llama-3\",   # or \"chatml\", \"mistral\", etc.\n)\n\nresp = llm.create_chat_completion(\n    messages=[\n        {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n        {\"role\": \"user\", \"content\": \"What is Python?\"},\n    ],\n    max_tokens=256,\n)\nprint(resp[\"choices\"][0][\"message\"][\"content\"])\n\n# Streaming\nfor chunk in llm(\"Explain quantum computing:\", max_tokens=256, stream=True):\n    print(chunk[\"choices\"][0][\"text\"], end=\"\", flush=True)\n```\n\n### Embeddings\n\n```python\nllm = Llama(model_path=\".\u002Fmodel-q4_k_m.gguf\", embedding=True, n_gpu_layers=35)\nvec = llm.embed(\"This is a test sentence.\")\nprint(f\"Embedding dimension: {len(vec)}\")\n```\n\nYou can also load a GGUF straight from the Hub:\n\n```python\nllm = Llama.from_pretrained(\n    repo_id=\"bartowski\u002FLlama-3.2-3B-Instruct-GGUF\",\n    filename=\"*Q4_K_M.gguf\",\n    n_gpu_layers=35,\n)\n```\n\n## Choosing a quant\n\nUse the Hub page first, generic heuristics second.\n\n- Prefer the exact quant that HF marks as compatible for the user's hardware profile.\n- For general chat, start with `Q4_K_M`.\n- For code or technical work, prefer `Q5_K_M` or `Q6_K` if memory allows.\n- For very tight RAM budgets, consider `Q3_K_M`, `IQ` variants, or `Q2` variants only if the user explicitly prioritizes fit over quality.\n- For multimodal repos, mention `mmproj-*.gguf` separately. The projector is not the main model file.\n- Do not normalize repo-native labels. If the page says `UD-Q4_K_M`, report `UD-Q4_K_M`.\n\n## Extracting available GGUFs from a repo\n\nWhen the user asks what GGUFs exist, return:\n\n- filename\n- file size\n- quant label\n- whether it is a main model or an auxiliary projector\n\nIgnore unless requested:\n\n- README\n- BF16 shard files\n- imatrix blobs or calibration artifacts\n\nUse the tree API for this step:\n\n- `https:\u002F\u002Fhuggingface.co\u002Fapi\u002Fmodels\u002F\u003Crepo>\u002Ftree\u002Fmain?recursive=true`\n\nFor a repo like `unsloth\u002FQwen3.6-35B-A3B-GGUF`, the local-app page can show quant chips such as `UD-Q4_K_M`, `UD-Q5_K_M`, `UD-Q6_K`, and `Q8_0`, while the tree API exposes exact file paths such as `Qwen3.6-35B-A3B-UD-Q4_K_M.gguf` and `Qwen3.6-35B-A3B-Q8_0.gguf` with byte sizes. Use the tree API to turn a quant label into an exact filename.\n\n## Search patterns\n\nUse these URL shapes directly:\n\n```text\nhttps:\u002F\u002Fhuggingface.co\u002Fmodels?apps=llama.cpp&sort=trending\nhttps:\u002F\u002Fhuggingface.co\u002Fmodels?search=\u003Cterm>&apps=llama.cpp&sort=trending\nhttps:\u002F\u002Fhuggingface.co\u002Fmodels?search=\u003Cterm>&apps=llama.cpp&num_parameters=min:0,max:24B&sort=trending\nhttps:\u002F\u002Fhuggingface.co\u002F\u003Crepo>?local-app=llama.cpp\nhttps:\u002F\u002Fhuggingface.co\u002Fapi\u002Fmodels\u002F\u003Crepo>\u002Ftree\u002Fmain?recursive=true\nhttps:\u002F\u002Fhuggingface.co\u002F\u003Crepo>\u002Ftree\u002Fmain\n```\n\n## Output format\n\nWhen answering discovery requests, prefer a compact structured result like:\n\n```text\nRepo: \u003Crepo>\nRecommended quant from HF: \u003Clabel> (\u003Csize>)\nllama-server: \u003Ccommand>\nOther GGUFs:\n- \u003Cfilename> - \u003Csize>\n- \u003Cfilename> - \u003Csize>\nSource URLs:\n- \u003Clocal-app URL>\n- \u003Ctree API URL>\n```\n\n## References\n\n- **[hub-discovery.md](references\u002Fhub-discovery.md)** - URL-only Hugging Face workflows, search patterns, GGUF extraction, and command reconstruction\n- **[advanced-usage.md](references\u002Fadvanced-usage.md)** — speculative decoding, batched inference, grammar-constrained generation, LoRA, multi-GPU, custom builds, benchmark scripts\n- **[quantization.md](references\u002Fquantization.md)** — quant quality tradeoffs, when to use Q4\u002FQ5\u002FQ6\u002FIQ, model size scaling, imatrix\n- **[server.md](references\u002Fserver.md)** — direct-from-Hub server launch, OpenAI API endpoints, Docker deployment, NGINX load balancing, monitoring\n- **[optimization.md](references\u002Foptimization.md)** — CPU threading, BLAS, GPU offload heuristics, batch tuning, benchmarks\n- **[troubleshooting.md](references\u002Ftroubleshooting.md)** — install\u002Fconvert\u002Fquantize\u002Finference\u002Fserver issues, Apple Silicon, debugging\n\n## Resources\n\n- **GitHub**: https:\u002F\u002Fgithub.com\u002Fggml-org\u002Fllama.cpp\n- **Hugging Face GGUF + llama.cpp docs**: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fgguf-llamacpp\n- **Hugging Face Local Apps docs**: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fmain\u002Flocal-apps\n- **Hugging Face Local Agents docs**: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fagents-local\n- **Example local-app page**: https:\u002F\u002Fhuggingface.co\u002Funsloth\u002FQwen3.6-35B-A3B-GGUF?local-app=llama.cpp\n- **Example tree API**: https:\u002F\u002Fhuggingface.co\u002Fapi\u002Fmodels\u002Funsloth\u002FQwen3.6-35B-A3B-GGUF\u002Ftree\u002Fmain?recursive=true\n- **Example llama.cpp search**: https:\u002F\u002Fhuggingface.co\u002Fmodels?num_parameters=min:0,max:24B&apps=llama.cpp&sort=trending\n- **License**: MIT\n",{"name":5,"author":23,"license":14,"version":24,"metadata":25,"platforms":39,"description":7,"dependencies":43},"Orchestra Research","2.1.2",{"hermes":26},{"tags":27},[28,29,30,31,32,33,34,35,36,37,38],"llama.cpp","GGUF","Quantization","Hugging Face Hub","CPU Inference","Apple Silicon","Edge Deployment","AMD GPUs","Intel GPUs","NVIDIA","URL-first",[40,41,42],"linux","macos","windows",[44],"llama-cpp-python>=0.2.0","main","e0ec177d00f087cb04e90206b2f7d3aa2f5a1350","References\u002FAgent\u002FPlatforms\u002Fhermes-agent\u002Fskills\u002Fmlops\u002Finference\u002Fllama-cpp","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim","https:\u002F\u002Fgithub.com\u002Flzy18001500226","2026-08-31 23:29:00",{"source":8,"skillName":5,"commands":52,"prompt":62,"zipAvailable":63},[53,56,59],{"tool":54,"command":55},"npx","npx skills add lzy18001500226\u002FMoSim --skill llama-cpp",{"tool":57,"command":58},"bunx","bunx skills add lzy18001500226\u002FMoSim --skill llama-cpp",{"tool":60,"command":61},"pnpm","pnpm dlx skills add lzy18001500226\u002FMoSim --skill llama-cpp","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002FReferences\u002FAgent\u002FPlatforms\u002Fhermes-agent\u002Fskills\u002Fmlops\u002Finference\u002Fllama-cpp\u002FSKILL.md",true,[65,71,75,79,83,87],{"path":66,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":69,"sourceHtmlUrl":70},"references\u002Fadvanced-usage.md","REFERENCE","参考文档",10887,"https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002FReferences\u002FAgent\u002FPlatforms\u002Fhermes-agent\u002Fskills\u002Fmlops\u002Finference\u002Fllama-cpp\u002Freferences\u002Fadvanced-usage.md",{"path":72,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":73,"sourceHtmlUrl":74},"references\u002Fhub-discovery.md",5082,"https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002FReferences\u002FAgent\u002FPlatforms\u002Fhermes-agent\u002Fskills\u002Fmlops\u002Finference\u002Fllama-cpp\u002Freferences\u002Fhub-discovery.md",{"path":76,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":77,"sourceHtmlUrl":78},"references\u002Foptimization.md",1659,"https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002FReferences\u002FAgent\u002FPlatforms\u002Fhermes-agent\u002Fskills\u002Fmlops\u002Finference\u002Fllama-cpp\u002Freferences\u002Foptimization.md",{"path":80,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":81,"sourceHtmlUrl":82},"references\u002Fquantization.md",5888,"https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002FReferences\u002FAgent\u002FPlatforms\u002Fhermes-agent\u002Fskills\u002Fmlops\u002Finference\u002Fllama-cpp\u002Freferences\u002Fquantization.md",{"path":84,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":85,"sourceHtmlUrl":86},"references\u002Fserver.md",2915,"https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002FReferences\u002FAgent\u002FPlatforms\u002Fhermes-agent\u002Fskills\u002Fmlops\u002Finference\u002Fllama-cpp\u002Freferences\u002Fserver.md",{"path":88,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":89,"sourceHtmlUrl":90},"references\u002Ftroubleshooting.md",8904,"https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002FReferences\u002FAgent\u002FPlatforms\u002Fhermes-agent\u002Fskills\u002Fmlops\u002Finference\u002Fllama-cpp\u002Freferences\u002Ftroubleshooting.md",[92,102,110,118,126,134],{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":96,"licenseFileUrl":97,"sourceHtmlUrl":98,"categories":99,"keywords":100,"publishedAt":101},"01M78RK6JGRPTD3TDSMTQWDQ9P","skill-installer","Install Codex skills into $CODEX_HOME\u002Fskills from a curated list or a GitHub repo path. Use when a user asks to list installable skills, install a curated skill, or install a skill from another repo (including private repos).","Apache-2.0","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002Fsrc\u002FAgent\u002Fcodex-main\u002Fcodex-rs\u002Fskills\u002Fsrc\u002Fassets\u002Fsamples\u002Fskill-installer\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002Fsrc\u002FAgent\u002Fcodex-main\u002Fcodex-rs\u002Fskills\u002Fsrc\u002Fassets\u002Fsamples\u002Fskill-installer\u002FSKILL.md",[],[],"2026-11-15 14:19:17",{"skillNo":103,"name":104,"nameZh":6,"description":105,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":96,"licenseFileUrl":106,"sourceHtmlUrl":107,"categories":108,"keywords":109,"publishedAt":101},"01M78RK29KVRDQXSMVZX1GQTJN","skill-creator","Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Codex's capabilities with specialized knowledge, workflows, or tool integrations.","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002Fsrc\u002FAgent\u002Fcodex-main\u002Fcodex-rs\u002Fskills\u002Fsrc\u002Fassets\u002Fsamples\u002Fskill-creator\u002Flicense.txt","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002Fsrc\u002FAgent\u002Fcodex-main\u002Fcodex-rs\u002Fskills\u002Fsrc\u002Fassets\u002Fsamples\u002Fskill-creator\u002FSKILL.md",[],[],{"skillNo":111,"name":112,"nameZh":6,"description":113,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":96,"licenseFileUrl":114,"sourceHtmlUrl":115,"categories":116,"keywords":117,"publishedAt":101},"01M78RJPF3210Y5DA9VJ6JMMG6","openai-docs","Use for Codex models\u002Fpricing, scheduled tasks, skills, settings, setup, troubleshooting, customization, automations, and self-knowledge—including 'you,' 'your,' 'this app,' or 'this coding agent' when they refer to Codex—and for OpenAI APIs\u002Fproducts and ChatGPT Work. Also use for model choice\u002Fmigration, prompting, SDKs, Responses, Realtime, agents, evals, and Chat\u002FWork\u002FCodex comparisons. Do not use for generic app\u002Fsoftware tasks that merely mention Codex.","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002Fsrc\u002FAgent\u002Fcodex-main\u002Fcodex-rs\u002Fskills\u002Fsrc\u002Fassets\u002Fsamples\u002Fopenai-docs\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002Fsrc\u002FAgent\u002Fcodex-main\u002Fcodex-rs\u002Fskills\u002Fsrc\u002Fassets\u002Fsamples\u002Fopenai-docs\u002FSKILL.md",[],[],{"skillNo":119,"name":120,"nameZh":6,"description":121,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":96,"licenseFileUrl":122,"sourceHtmlUrl":123,"categories":124,"keywords":125,"publishedAt":101},"01M78RJH7X67C5AY8ZVS7RQNWJ","imagegen","Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output should be a bitmap asset rather than repo-native code or vector. Do not use when the task is better handled by editing existing SVG\u002Fvector\u002Fcode-native assets, extending an established icon or logo system, or building the visual directly in HTML\u002FCSS\u002Fcanvas.","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002Fsrc\u002FAgent\u002Fcodex-main\u002Fcodex-rs\u002Fskills\u002Fsrc\u002Fassets\u002Fsamples\u002Fimagegen\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002Fsrc\u002FAgent\u002Fcodex-main\u002Fcodex-rs\u002Fskills\u002Fsrc\u002Fassets\u002Fsamples\u002Fimagegen\u002FSKILL.md",[],[],{"skillNo":127,"name":128,"nameZh":6,"description":129,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":96,"licenseFileUrl":130,"sourceHtmlUrl":131,"categories":132,"keywords":133,"publishedAt":101},"01M78RBBAYMJ6T3ZMAYHCFM6G7","skill-creator-cc","Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, update or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002FReferences\u002FAgent\u002FWorkflow\u002Fokwinds\u002Fmiscellany\u002Fagent\u002Fskills\u002Fskill-creator-cc\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002FReferences\u002FAgent\u002FWorkflow\u002Fokwinds\u002Fmiscellany\u002Fagent\u002Fskills\u002Fskill-creator-cc\u002FSKILL.md",[],[],{"skillNo":135,"name":136,"nameZh":6,"description":137,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":138,"sourceHtmlUrl":139,"categories":140,"keywords":141,"publishedAt":142},"01M76VV9V2R9FZTVGMJZV7Y996","prd-writing-guide","Write complete, unambiguous PRDs that development teams can implement without guesswork. Includes requirement discovery framework, structured documentation methodology, completeness checklists, and common pitfall avoidance. Use when: writing new PRDs, reviewing PRD drafts, validating requirement completeness, preparing for engineering handoff. Triggers: 'write PRD', '写PRD', '产品需求文档', '需求文档', '需求规格', '需求评审', '完善需求', 'create requirements doc', 'product requirements', 'feature spec', 'requirements document'. Anti-triggers: 'technical design doc', 'architecture design', 'implementation plan', 'API design', '架构设计', '技术方案', '实现方案', '接口设计'.","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002FReferences\u002FAgent\u002FWorkflow\u002Fokwinds\u002Fmiscellany\u002Fagent\u002Fskills\u002Fprd-writing-guide\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Flzy18001500226\u002FMoSim\u002Fblob\u002Fmain\u002FReferences\u002FAgent\u002FWorkflow\u002Fokwinds\u002Fmiscellany\u002Fagent\u002Fskills\u002Fprd-writing-guide\u002FSKILL.md",[],[],"2026-11-12 13:55:43"]