[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78WWP259YP6Q0N2CB7NEPKT":3,"skill-same:01M78WWP259YP6Q0N2CB7NEPKT":66},{"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":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":32,"install":33,"resources":45},"01M78WWP259YP6Q0N2CB7NEPKT","accelerated-and-interoperable",null,"This skill guides Faiss CPU\u002FGPU acceleration, backend-gated interoperability, native C\u002FC++ APIs, Torch buffers, and safe RPC boundaries.","VectorSpaceLab\u002FAREX-Skill","VectorSpaceLab","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F180637102?v=4","01M7538C5WPSJGWF7KHR4EFZ9F","01M7538C6R2GFP53NPXTCZNBRT",351,"MIT","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ffaiss\u002Fsub-skills\u002Faccelerated-and-interoperable\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: accelerated-and-interoperable\ndescription: \"This skill guides Faiss CPU\u002FGPU acceleration, backend-gated\n  interoperability, native C\u002FC++ APIs, Torch buffers, and safe RPC boundaries.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: MIT\n---\n\n# Accelerated and Interoperable Faiss\n\nUse this sub-skill when a task names `faiss-gpu`, `StandardGpuResources`,\n`index_cpu_to_gpu`, `index_cpu_to_all_gpus`, CUDA or multi-GPU execution, cuVS,\nROCm, Metal, SVS, CMake\u002FC++20, the C API, Torch tensors, FFI, or distributed\nFaiss\u002FRPC boundaries.\n\n## Route here for\n\n- Checking whether the **installed** Faiss binding was built with GPU, cuVS,\n  SVS, or Metal support; distinguishing package\u002Fbuild support from visible\n  hardware.\n- Cloning a CPU index to one GPU or multiple GPUs, selecting replication versus\n  IVF sharding, and copying GPU results back for parity checks.\n- Building the optional CUDA, ROCm, cuVS, Metal, or SVS backends, with explicit\n  prerequisites and no claims that an unverified backend works.\n- C++20\u002FCMake integration, C API\u002FFFI naming and ownership, Torch tensor\n  pointers, CUDA device\u002Fstream ordering, and dynamic-library\u002FABI diagnosis.\n- Treating the repository's demonstration RPC as a trusted-network example,\n  not an authenticated production service.\n\n## Route away\n\n- Generic CPU index construction, search parameters, and ordinary NumPy\n  recipes: use [index-selection-and-search](..\u002Findex-selection-and-search\u002FSKILL.md).\n- Codec, quantizer, memory-layout, or precision tuning: use\n  [training-and-compression](..\u002Ftraining-and-compression\u002FSKILL.md).\n- CPU composition, ID selectors, and ordinary shards\u002Freplicas: use\n  [composition-and-filtering](..\u002Fcomposition-and-filtering\u002FSKILL.md).\n- Persistence, serialization, and evaluation methodology: use\n  [persistence-and-evaluation](..\u002Fpersistence-and-evaluation\u002FSKILL.md).\n- Benchmarks, CI, release engineering, and large benchmark datasets are out of\n  scope here; use the appropriate sibling or project workflow.\n\n## Operating order\n\n1. Run [the backend checker](scripts\u002Fcheck_backend.py) in the target runtime.\n   Treat `import faiss` failure, absent GPU symbols, and `get_num_gpus() == 0`\n   as distinct signals.\n2. Select the matrix row and gate in [the API reference](references\u002Fapi-reference.md).\n   The prepared inspection facts are CPU-only: `faiss-cpu 1.15.0`, Python\n   >=3.10, NumPy `float32`, compile options `OPTIMIZE DD AVX2`, and zero\n   reported GPUs. CUDA hardware on the host is not evidence that a CUDA Faiss\n   package or runtime is usable.\n3. For a GPU task, prove both package symbols and a visible device before\n   allocating `StandardGpuResources`. If either gate fails, use the CPU\n   fallback; do not install, mutate, or silently select a different backend.\n4. Start with a small exact `IndexFlatL2` CPU baseline, clone only after the\n   GPU gate passes, and compare labels and distances after copying results to\n   host memory. Record tolerance and device IDs; parity is workload- and\n   backend-dependent, not a universal guarantee.\n5. For native integration, match C++ standard, compiler ABI, CUDA\u002FROCm toolkit,\n   BLAS, Python\u002FNumPy ABI, and shared-library search paths. Keep C API object\n   ownership and error-code checks explicit.\n6. For Torch, keep tensors contiguous, use compatible dtype\u002Fdevice pairs, and\n   order Faiss work with the current CUDA stream. Never pass a CUDA tensor to a\n   CPU index or mix NumPy arrays and Torch tensors in one patched call.\n7. Treat RPC as an explicit trust boundary. The bundled pickle protocol has no\n   authentication or transport encryption and is not safe on an untrusted\n   network; isolate it or put an authenticated, encrypted, allow-listed service\n   in front of it.\n\n## Bundled resources\n\n- [API and compatibility reference](references\u002Fapi-reference.md) — package and\n  source-build matrix, probes, public GPU\u002FC++\u002FC\u002FTorch\u002FRPC contracts, and ABI\n  notes.\n- [Workflows](references\u002Fworkflows.md) — safe backend inspection, CPU fallback,\n  tiny one-\u002Fmulti-GPU transfer\u002Fparity, CMake\u002FC API, Torch, and RPC boundary\n  workflows.\n- [Troubleshooting](references\u002Ftroubleshooting.md) — CPU-build AttributeError,\n  missing toolkit\u002Fbackend gates, loader\u002FABI failures, device\u002Fstream errors, and\n  RPC safety failures.\n- [Backend checker](scripts\u002Fcheck_backend.py) — deterministic, no-allocation\n  availability report that is safe to run from any working directory and\n  gracefully handles missing optional dependencies.\n\n## Verification stance\n\nThe CPU package and CPU smoke path were verified. GPU, cuVS, ROCm, Metal, SVS,\nTorch-GPU, and network\u002FRPC runtime behavior remain conditional until the target\npackage, build flags, dynamic libraries, device visibility, and a focused smoke\ncheck all pass in the target environment.\n","# Accelerated and Interoperable Faiss\n\nUse this sub-skill when a task names `faiss-gpu`, `StandardGpuResources`,\n`index_cpu_to_gpu`, `index_cpu_to_all_gpus`, CUDA or multi-GPU execution, cuVS,\nROCm, Metal, SVS, CMake\u002FC++20, the C API, Torch tensors, FFI, or distributed\nFaiss\u002FRPC boundaries.\n\n## Route here for\n\n- Checking whether the **installed** Faiss binding was built with GPU, cuVS,\n  SVS, or Metal support; distinguishing package\u002Fbuild support from visible\n  hardware.\n- Cloning a CPU index to one GPU or multiple GPUs, selecting replication versus\n  IVF sharding, and copying GPU results back for parity checks.\n- Building the optional CUDA, ROCm, cuVS, Metal, or SVS backends, with explicit\n  prerequisites and no claims that an unverified backend works.\n- C++20\u002FCMake integration, C API\u002FFFI naming and ownership, Torch tensor\n  pointers, CUDA device\u002Fstream ordering, and dynamic-library\u002FABI diagnosis.\n- Treating the repository's demonstration RPC as a trusted-network example,\n  not an authenticated production service.\n\n## Route away\n\n- Generic CPU index construction, search parameters, and ordinary NumPy\n  recipes: use [index-selection-and-search](..\u002Findex-selection-and-search\u002FSKILL.md).\n- Codec, quantizer, memory-layout, or precision tuning: use\n  [training-and-compression](..\u002Ftraining-and-compression\u002FSKILL.md).\n- CPU composition, ID selectors, and ordinary shards\u002Freplicas: use\n  [composition-and-filtering](..\u002Fcomposition-and-filtering\u002FSKILL.md).\n- Persistence, serialization, and evaluation methodology: use\n  [persistence-and-evaluation](..\u002Fpersistence-and-evaluation\u002FSKILL.md).\n- Benchmarks, CI, release engineering, and large benchmark datasets are out of\n  scope here; use the appropriate sibling or project workflow.\n\n## Operating order\n\n1. Run [the backend checker](scripts\u002Fcheck_backend.py) in the target runtime.\n   Treat `import faiss` failure, absent GPU symbols, and `get_num_gpus() == 0`\n   as distinct signals.\n2. Select the matrix row and gate in [the API reference](references\u002Fapi-reference.md).\n   The prepared inspection facts are CPU-only: `faiss-cpu 1.15.0`, Python\n   >=3.10, NumPy `float32`, compile options `OPTIMIZE DD AVX2`, and zero\n   reported GPUs. CUDA hardware on the host is not evidence that a CUDA Faiss\n   package or runtime is usable.\n3. For a GPU task, prove both package symbols and a visible device before\n   allocating `StandardGpuResources`. If either gate fails, use the CPU\n   fallback; do not install, mutate, or silently select a different backend.\n4. Start with a small exact `IndexFlatL2` CPU baseline, clone only after the\n   GPU gate passes, and compare labels and distances after copying results to\n   host memory. Record tolerance and device IDs; parity is workload- and\n   backend-dependent, not a universal guarantee.\n5. For native integration, match C++ standard, compiler ABI, CUDA\u002FROCm toolkit,\n   BLAS, Python\u002FNumPy ABI, and shared-library search paths. Keep C API object\n   ownership and error-code checks explicit.\n6. For Torch, keep tensors contiguous, use compatible dtype\u002Fdevice pairs, and\n   order Faiss work with the current CUDA stream. Never pass a CUDA tensor to a\n   CPU index or mix NumPy arrays and Torch tensors in one patched call.\n7. Treat RPC as an explicit trust boundary. The bundled pickle protocol has no\n   authentication or transport encryption and is not safe on an untrusted\n   network; isolate it or put an authenticated, encrypted, allow-listed service\n   in front of it.\n\n## Bundled resources\n\n- [API and compatibility reference](references\u002Fapi-reference.md) — package and\n  source-build matrix, probes, public GPU\u002FC++\u002FC\u002FTorch\u002FRPC contracts, and ABI\n  notes.\n- [Workflows](references\u002Fworkflows.md) — safe backend inspection, CPU fallback,\n  tiny one-\u002Fmulti-GPU transfer\u002Fparity, CMake\u002FC API, Torch, and RPC boundary\n  workflows.\n- [Troubleshooting](references\u002Ftroubleshooting.md) — CPU-build AttributeError,\n  missing toolkit\u002Fbackend gates, loader\u002FABI failures, device\u002Fstream errors, and\n  RPC safety failures.\n- [Backend checker](scripts\u002Fcheck_backend.py) — deterministic, no-allocation\n  availability report that is safe to run from any working directory and\n  gracefully handles missing optional dependencies.\n\n## Verification stance\n\nThe CPU package and CPU smoke path were verified. GPU, cuVS, ROCm, Metal, SVS,\nTorch-GPU, and network\u002FRPC runtime behavior remain conditional until the target\npackage, build flags, dynamic libraries, device visibility, and a focused smoke\ncheck all pass in the target environment.\n",{"name":5,"license":14,"metadata":23,"description":7,"disable-model-invocation":25},{"disco-role":24},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Ffaiss\u002Fsub-skills\u002Faccelerated-and-interoperable",27,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab","2026-09-03 10:25:21",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":25},[35,38,41],{"tool":36,"command":37},"npx","npx skills add VectorSpaceLab\u002FAREX-Skill --skill accelerated-and-interoperable",{"tool":39,"command":40},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill accelerated-and-interoperable",{"tool":42,"command":43},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill accelerated-and-interoperable","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ffaiss\u002Fsub-skills\u002Faccelerated-and-interoperable\u002FSKILL.md",[46,52,56,60],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"references\u002Fapi-reference.md","REFERENCE","参考文档",10878,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ffaiss\u002Fsub-skills\u002Faccelerated-and-interoperable\u002Freferences\u002Fapi-reference.md",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Ftroubleshooting.md",8478,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ffaiss\u002Fsub-skills\u002Faccelerated-and-interoperable\u002Freferences\u002Ftroubleshooting.md",{"path":57,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":58,"sourceHtmlUrl":59},"references\u002Fworkflows.md",9393,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ffaiss\u002Fsub-skills\u002Faccelerated-and-interoperable\u002Freferences\u002Fworkflows.md",{"path":61,"resourceType":62,"resourceTypeDesc":63,"sizeBytes":64,"sourceHtmlUrl":65},"scripts\u002Fcheck_backend.py","SCRIPT","脚本",9237,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Ffaiss\u002Fsub-skills\u002Faccelerated-and-interoperable\u002Fscripts\u002Fcheck_backend.py",[67,75,82,89,96,103],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":71,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":18},"01M799NYSH1R3RVAHSM2D88Y84","passnet-triton-opt","Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block\u002Fgrid\u002Fwarp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.","Apache-2.0","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-triton-opt\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":71,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":18},"01M799NS5APEC0N1FZ7XZVW6BN","passnet-solve","END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-solve\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":71,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":18},"01M799NNY9BKSM21V5R7QVTBJX","passnet-skill","PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under .\u002Fpass_dir\u002F, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-skill\u002FSKILL.md",[],[],{"skillNo":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":71,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"publishedAt":18},"01M799NHH6NP7Z1RK77KGXAN6S","passnet-pattern-fusion","Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()\u002Freplacement_args()\u002Freplacement_func() files, when a pass \"failed to match\", or when deciding how to fuse multiple kernels\u002Fops.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-pattern-fusion\u002FSKILL.md",[],[],{"skillNo":97,"name":98,"nameZh":6,"description":99,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":71,"licenseFileUrl":6,"sourceHtmlUrl":100,"categories":101,"keywords":102,"publishedAt":18},"01M799ND2J4WNC2PGAZ0A06NDF","passnet-orchestrate","Round\u002Fbatch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep\u002Frevert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-orchestrate\u002FSKILL.md",[],[],{"skillNo":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":71,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"publishedAt":18},"01M799N8SBTMY02ZDG2AZF5ZWE","passnet-feedback","Fast, deterministic feedback for PassNet work: pre-flight pattern\u002Fmatch verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).","https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Ftask-oriented\u002FPassNet\u002Fskills\u002Fpassnet-feedback\u002FSKILL.md",[],[]]