[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M794T97DMN3PMNG4SM3KH742":3,"skill-same:01M794T97DMN3PMNG4SM3KH742":64},{"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},"01M794T97DMN3PMNG4SM3KH742","second-pytorch",null,"Guide historical SECOND and PointPillars LiDAR 3D detection workflows for KITTI and NuScenes, including data contracts, configuration, geometry, evaluation, guarded training\u002Finference, and the web viewer.","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\u002Fsecond-pytorch\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: second-pytorch\ndescription: \"Guide historical SECOND and PointPillars LiDAR 3D detection\n  workflows for KITTI and NuScenes, including data contracts, configuration,\n  geometry, evaluation, guarded training\u002Finference, and the web viewer.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: MIT\n---\n\n# SECOND PyTorch\n\nUse this repo skill for source-grounded work involving the deprecated\n`traveller59\u002Fsecond.pytorch` 1.6.0-alpha-era codebase: KITTI\u002FNuScenes LiDAR\n3D detection, VoxelNet\u002FSECOND, PointPillars, voxel and box geometry, text\nprotobuf configs, evaluation, and the historical KITTI web viewer.\n\n## First decision: historical code or maintained replacement?\n\nThe repository is explicitly deprecated. For new detector development, prefer\nOpenPCDet or MMDetection3D. Use this skill when a historical config, checkpoint,\nexperiment, data artifact, or compatibility question requires this code's\ncontracts. Read [repo-provenance.md](references\u002Frepo-provenance.md) before\nassuming this skill matches a checkout, and read\n[compatibility.md](references\u002Fcompatibility.md) before importing the detector.\n\n## Runtime boundary\n\n- The checkout has no `pyproject.toml`, `setup.py`, requirements file, or\n  supported modern package install. Use a private isolated environment and an\n  explicitly supplied checkout only when the historical runtime is truly needed.\n- The executable, verified portion of this graph is CPU-safe\u002Fstatic: dataset\n  layout validation, config\u002Fschema reasoning, NumPy geometry, and safe helper\n  probes.\n- Full sparse detector construction, GPU NMS, training, evaluation, checkpoint\n  restoration, and viewer checkpoint inference are **legacy-backend guarded**.\n  Current `spconv` 2.x APIs are not interchangeable with the source's old\n  symbols. A successful Torch CUDA smoke or `spconv` import is not detector\n  verification.\n- Do not download datasets\u002Fcheckpoints, start a service, launch the Qt viewer,\n  run full training\u002Fevaluation, or mutate data as a skill smoke test.\n\nRead [troubleshooting.md](references\u002Ftroubleshooting.md) for cross-cutting\nfailures before changing dependencies, paths, or config files.\n\n## Public environment setup\n\nThere is no supported package installer for this source tree. For the verified\nCPU-safe\u002Fstatic routes, create an isolated Python 3.10+ environment and install\nonly the public scientific\u002Fruntime dependencies needed by the selected helper,\nfor example `numpy`, `scipy`, `numba`, `scikit-image`, `Pillow`, `protobuf\u003C4`,\n`fire`, and optional `nuscenes-devkit` or `flask-cors`. The bundled geometry and\nlayout helpers use standard Python\u002FNumPy and do not require the detector stack.\n\nFor model work, install a separately reproducible legacy Torch\u002Fspconv\u002FNumba\ncombination and run the backend probe. Do not install the newest `spconv` and\nassume compatibility; the source expects old symbols that current 2.x wheels\nmay not expose. Never mutate an existing environment or use a CUDA import as a\nsubstitute for the gate in [compatibility.md](references\u002Fcompatibility.md).\n\n## Route by task\n\n| Request signal | Read next |\n|---|---|\n| KITTI\u002FNuScenes root, infos, reduced clouds, ground-truth database, sweeps, custom dataset, voxel\u002Fpreprocess layout | [data-preparation](sub-skills\u002Fdata-preparation\u002FSKILL.md) |\n| train\u002Fevaluate, VoxelNet, PointPillars, checkpoint, config selection, fp16, multi-GPU, spconv\u002FNumba compatibility | [training-and-inference](sub-skills\u002Ftraining-and-inference\u002FSKILL.md) |\n| boxes, corners, yaw, camera\u002Flidar conversion, anchors, encoding, IoU\u002FNMS, KITTI AP, NuScenes results | [geometry-and-evaluation](sub-skills\u002Fgeometry-and-evaluation\u002FSKILL.md) |\n| KITTI web viewer, Flask\u002FCORS, browser\u002Fbackend URLs, point-cloud or detection payloads, buildNet\u002Finference UI | [visualization-and-serving](sub-skills\u002Fvisualization-and-serving\u002FSKILL.md) |\n\nCross-route requests should start with data preparation, then configuration and\ntraining\u002Finference, while geometry\u002Fevaluation owns box conventions and metrics.\nThe viewer is a guarded consumer of the same data and checkpoint contracts.\n\n## Minimal safe checks\n\nRun the bundled helpers from their linked sub-skills before any user-controlled\nworkflow. They are self-contained and do not require the original checkout:\n\n```bash\npython sub-skills\u002Fdata-preparation\u002Fscripts\u002Fvalidate_dataset_layout.py --help\npython sub-skills\u002Fgeometry-and-evaluation\u002Fscripts\u002Fgeometry_smoke.py\npython sub-skills\u002Ftraining-and-inference\u002Fscripts\u002Fcheck_legacy_backend.py\npython sub-skills\u002Fvisualization-and-serving\u002Fscripts\u002Fcheck_viewer_deps.py --json\n```\n\nThe geometry helper should finish with `geometry smoke: PASS`. The other probes\nare diagnostic; missing optional components or legacy symbols must be reported,\nnot silently substituted.\n\n## Public contract reminders\n\n- Internal LiDAR boxes normally use `[x, y, z, w, l, h, yaw]`; KITTI camera\n  boxes use a different dimension\u002Forder convention. Never swap dimensions by\n  visual intuition.\n- KITTI preparation expects `training\u002F` and `testing\u002F` trees with matching\n  image, calibration, point-cloud, and (for training) label stems. NuScenes\n  preparation depends on version metadata, samples, sweeps, and a matching\n  velocity\u002Fnon-velocity dataset class.\n- Historical config field names such as `kitti_root_path` and\n  `kitti_info_path` are used by both dataset families. Keep dataset class,\n  info files, feature width, box dimensionality, and class order consistent.\n\n## Handoff\n\nA useful result reports the selected route, source\u002Fskill provenance, validated\nlayout\u002Fconfig\u002Fgeometry observations, exact generated output paths, backend gate\nstatus, and any unverified legacy limitation. Do not claim a metric, detector\nrun, or viewer inference result from a static or CPU-safe check alone.\n","# SECOND PyTorch\n\nUse this repo skill for source-grounded work involving the deprecated\n`traveller59\u002Fsecond.pytorch` 1.6.0-alpha-era codebase: KITTI\u002FNuScenes LiDAR\n3D detection, VoxelNet\u002FSECOND, PointPillars, voxel and box geometry, text\nprotobuf configs, evaluation, and the historical KITTI web viewer.\n\n## First decision: historical code or maintained replacement?\n\nThe repository is explicitly deprecated. For new detector development, prefer\nOpenPCDet or MMDetection3D. Use this skill when a historical config, checkpoint,\nexperiment, data artifact, or compatibility question requires this code's\ncontracts. Read [repo-provenance.md](references\u002Frepo-provenance.md) before\nassuming this skill matches a checkout, and read\n[compatibility.md](references\u002Fcompatibility.md) before importing the detector.\n\n## Runtime boundary\n\n- The checkout has no `pyproject.toml`, `setup.py`, requirements file, or\n  supported modern package install. Use a private isolated environment and an\n  explicitly supplied checkout only when the historical runtime is truly needed.\n- The executable, verified portion of this graph is CPU-safe\u002Fstatic: dataset\n  layout validation, config\u002Fschema reasoning, NumPy geometry, and safe helper\n  probes.\n- Full sparse detector construction, GPU NMS, training, evaluation, checkpoint\n  restoration, and viewer checkpoint inference are **legacy-backend guarded**.\n  Current `spconv` 2.x APIs are not interchangeable with the source's old\n  symbols. A successful Torch CUDA smoke or `spconv` import is not detector\n  verification.\n- Do not download datasets\u002Fcheckpoints, start a service, launch the Qt viewer,\n  run full training\u002Fevaluation, or mutate data as a skill smoke test.\n\nRead [troubleshooting.md](references\u002Ftroubleshooting.md) for cross-cutting\nfailures before changing dependencies, paths, or config files.\n\n## Public environment setup\n\nThere is no supported package installer for this source tree. For the verified\nCPU-safe\u002Fstatic routes, create an isolated Python 3.10+ environment and install\nonly the public scientific\u002Fruntime dependencies needed by the selected helper,\nfor example `numpy`, `scipy`, `numba`, `scikit-image`, `Pillow`, `protobuf\u003C4`,\n`fire`, and optional `nuscenes-devkit` or `flask-cors`. The bundled geometry and\nlayout helpers use standard Python\u002FNumPy and do not require the detector stack.\n\nFor model work, install a separately reproducible legacy Torch\u002Fspconv\u002FNumba\ncombination and run the backend probe. Do not install the newest `spconv` and\nassume compatibility; the source expects old symbols that current 2.x wheels\nmay not expose. Never mutate an existing environment or use a CUDA import as a\nsubstitute for the gate in [compatibility.md](references\u002Fcompatibility.md).\n\n## Route by task\n\n| Request signal | Read next |\n|---|---|\n| KITTI\u002FNuScenes root, infos, reduced clouds, ground-truth database, sweeps, custom dataset, voxel\u002Fpreprocess layout | [data-preparation](sub-skills\u002Fdata-preparation\u002FSKILL.md) |\n| train\u002Fevaluate, VoxelNet, PointPillars, checkpoint, config selection, fp16, multi-GPU, spconv\u002FNumba compatibility | [training-and-inference](sub-skills\u002Ftraining-and-inference\u002FSKILL.md) |\n| boxes, corners, yaw, camera\u002Flidar conversion, anchors, encoding, IoU\u002FNMS, KITTI AP, NuScenes results | [geometry-and-evaluation](sub-skills\u002Fgeometry-and-evaluation\u002FSKILL.md) |\n| KITTI web viewer, Flask\u002FCORS, browser\u002Fbackend URLs, point-cloud or detection payloads, buildNet\u002Finference UI | [visualization-and-serving](sub-skills\u002Fvisualization-and-serving\u002FSKILL.md) |\n\nCross-route requests should start with data preparation, then configuration and\ntraining\u002Finference, while geometry\u002Fevaluation owns box conventions and metrics.\nThe viewer is a guarded consumer of the same data and checkpoint contracts.\n\n## Minimal safe checks\n\nRun the bundled helpers from their linked sub-skills before any user-controlled\nworkflow. They are self-contained and do not require the original checkout:\n\n```bash\npython sub-skills\u002Fdata-preparation\u002Fscripts\u002Fvalidate_dataset_layout.py --help\npython sub-skills\u002Fgeometry-and-evaluation\u002Fscripts\u002Fgeometry_smoke.py\npython sub-skills\u002Ftraining-and-inference\u002Fscripts\u002Fcheck_legacy_backend.py\npython sub-skills\u002Fvisualization-and-serving\u002Fscripts\u002Fcheck_viewer_deps.py --json\n```\n\nThe geometry helper should finish with `geometry smoke: PASS`. The other probes\nare diagnostic; missing optional components or legacy symbols must be reported,\nnot silently substituted.\n\n## Public contract reminders\n\n- Internal LiDAR boxes normally use `[x, y, z, w, l, h, yaw]`; KITTI camera\n  boxes use a different dimension\u002Forder convention. Never swap dimensions by\n  visual intuition.\n- KITTI preparation expects `training\u002F` and `testing\u002F` trees with matching\n  image, calibration, point-cloud, and (for training) label stems. NuScenes\n  preparation depends on version metadata, samples, sweeps, and a matching\n  velocity\u002Fnon-velocity dataset class.\n- Historical config field names such as `kitti_root_path` and\n  `kitti_info_path` are used by both dataset families. Keep dataset class,\n  info files, feature width, box dimensionality, and class order consistent.\n\n## Handoff\n\nA useful result reports the selected route, source\u002Fskill provenance, validated\nlayout\u002Fconfig\u002Fgeometry observations, exact generated output paths, backend gate\nstatus, and any unverified legacy limitation. Do not claim a metric, detector\nrun, or viewer inference result from a static or CPU-safe check alone.\n",{"name":5,"license":14,"metadata":23,"description":7,"disable-model-invocation":25},{"disco-role":24},"operating",true,"main","ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6","skills\u002Frepositories\u002Frepo-skills\u002Fsecond-pytorch",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 second-pytorch",{"tool":39,"command":40},"bunx","bunx skills add VectorSpaceLab\u002FAREX-Skill --skill second-pytorch",{"tool":42,"command":43},"pnpm","pnpm dlx skills add VectorSpaceLab\u002FAREX-Skill --skill second-pytorch","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fsecond-pytorch\u002FSKILL.md",[46,52,56,60],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"references\u002Fcompatibility.md","REFERENCE","参考文档",3272,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fsecond-pytorch\u002Freferences\u002Fcompatibility.md",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Frepo-provenance.md",2144,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fsecond-pytorch\u002Freferences\u002Frepo-provenance.md",{"path":57,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":58,"sourceHtmlUrl":59},"references\u002Frepo-routing-metadata.json",337,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fsecond-pytorch\u002Freferences\u002Frepo-routing-metadata.json",{"path":61,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":62,"sourceHtmlUrl":63},"references\u002Ftroubleshooting.md",4863,"https:\u002F\u002Fgithub.com\u002FVectorSpaceLab\u002FAREX-Skill\u002Fblob\u002Fmain\u002Fskills\u002Frepositories\u002Frepo-skills\u002Fsecond-pytorch\u002Freferences\u002Ftroubleshooting.md",[65,73,80,87,94,101],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":69,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":72,"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":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":69,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"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":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":69,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"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":88,"name":89,"nameZh":6,"description":90,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":69,"licenseFileUrl":6,"sourceHtmlUrl":91,"categories":92,"keywords":93,"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":95,"name":96,"nameZh":6,"description":97,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":69,"licenseFileUrl":6,"sourceHtmlUrl":98,"categories":99,"keywords":100,"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":102,"name":103,"nameZh":6,"description":104,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":69,"licenseFileUrl":6,"sourceHtmlUrl":105,"categories":106,"keywords":107,"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",[],[]]