[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7F9V14YA6DK3RP7E79WVC0Z":3,"skill-same:01M7F9V14YA6DK3RP7E79WVC0Z":63},{"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":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":28,"commitSha":29,"skillDirectory":30,"forksCount":31,"repositoryHtmlUrl":32,"repositoryHomepageUrl":6,"ownerHtmlUrl":33,"ownerBlogUrl":6,"ownerDescription":34,"updatedAt":35,"install":36,"resources":48},"01M7F9V14YA6DK3RP7E79WVC0Z","data-preparation",null,"Prepare, validate, and visualize HRM ARC, Sudoku, and Maze puzzle datasets in the repository's shared NumPy layout.","lza6\u002FArmor-piercing-system-prompts","lza6","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F128880206?v=4","01M7EB3K7WPX5T28EJ2NVHE6EG","Chinese-tingfeng","01M7EB3K97E1KQ16QWKJMFNGX8",11,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F06-前端与设计\u002Fhrm\u002Fsub-skills\u002Fdata-preparation\u002FSKILL.md",[],[],"2026-11-16 10:05:34",false,"---\nname: data-preparation\ndescription: \"Prepare, validate, and visualize HRM ARC, Sudoku, and Maze puzzle\n  datasets in the repository's shared NumPy layout.\"\ndisable-model-invocation: true\nmetadata:\n  disco-role: operating\nlicense: Apache 2.0\n---\n\n# HRM Data Preparation\n\nUse this sub-skill when the task is to create, inspect, validate, or visualize\nHierarchical Reasoning Model (HRM) datasets for ARC, Sudoku, or Maze workflows.\nThe original repo builders produce a shared directory layout consumed by\n`PuzzleDataset` and by `pretrain.py`\u002F`evaluate.py`.\n\n## When to use\n\n- The user asks to build ARC-AGI, ARC-AGI-2, ConceptARC, Sudoku Extreme, or Maze\n  datasets for HRM.\n- The task names `dataset\u002Fbuild_arc_dataset.py`, `build_sudoku_dataset.py`,\n  `build_maze_dataset.py`, `PuzzleDataset`, `PuzzleDatasetMetadata`,\n  `identifiers.json`, or `all__inputs.npy` style files.\n- A training or evaluation run fails before model execution because dataset\n  paths, metadata, `.npy` arrays, split\u002Fsubset names, token ranges, or puzzle\n  index arrays are wrong.\n- The user wants a browser visualizer for converted puzzle folders.\n\n## Route map\n\n1. Read [references\u002Fdata-formats.md](references\u002Fdata-formats.md) for the exact\n   shared dataset directory structure, metadata fields, array meanings, and\n   validation invariants.\n2. Read [references\u002Fworkflows.md](references\u002Fworkflows.md) for ARC, Sudoku,\n   Maze, and visualization command recipes, including network\u002Fsubmodule\n   prerequisites and safe bounded variants.\n3. Use [references\u002Ftroubleshooting.md](references\u002Ftroubleshooting.md) when a\n   builder cannot find raw data, Hugging Face downloads fail, token ranges look\n   invalid, or `PuzzleDataset` yields no batches.\n4. Run [scripts\u002Fvalidate_dataset_layout.py](scripts\u002Fvalidate_dataset_layout.py)\n   against a converted dataset before long training or evaluation.\n5. Use [scripts\u002Frun_dataset_builder.py](scripts\u002Frun_dataset_builder.py) when a\n   future task needs a wrapper that forwards arguments to an HRM checkout from\n   any current working directory.\n6. Open [scripts\u002Fpuzzle_visualizer.html](scripts\u002Fpuzzle_visualizer.html) in a\n   browser for local inspection of generated puzzle folders. The required\n   `npyjs` asset is bundled under `scripts\u002Fassets\u002F`.\n\n## Core workflow\n\n```bash\n# ARC-1 from initialized ARC-AGI + ConceptARC submodules\npython dataset\u002Fbuild_arc_dataset.py\n\n# ARC-2 from ARC-AGI-2 raw data\npython dataset\u002Fbuild_arc_dataset.py \\\n  --dataset-dirs dataset\u002Fraw-data\u002FARC-AGI-2\u002Fdata \\\n  --output-dir data\u002Farc-2-aug-1000\n\n# Sudoku 1K augmented demo\npython dataset\u002Fbuild_sudoku_dataset.py \\\n  --output-dir data\u002Fsudoku-extreme-1k-aug-1000 \\\n  --subsample-size 1000 \\\n  --num-aug 1000\n\n# Maze 30x30 hard 1K\npython dataset\u002Fbuild_maze_dataset.py --output-dir data\u002Fmaze-30x30-hard-1k\n```\n\nAfter building, validate the shared layout before training:\n\n```bash\npython \u003Cskill>\u002Fsub-skills\u002Fdata-preparation\u002Fscripts\u002Fvalidate_dataset_layout.py \\\n  data\u002Fsudoku-extreme-1k-aug-1000 --splits train test\n```\n\nReplace `\u003Cskill>` with the root of this generated skill directory. The helper is\nself-contained; it does not need the original HRM checkout unless you are\nrunning the original builder scripts.\n\n## Boundaries\n\n- This sub-skill owns dataset builders, converted layout, schema validation,\n  and local visualization.\n- Use `model-architecture` for HRM ACT v1 config fields, losses, model class\n  routing, FlashAttention layers, and sparse puzzle embeddings.\n- Use `training-evaluation` for `pretrain.py`, `evaluate.py`, checkpoints,\n  W&B, CUDA training, and ARC prediction post-processing.\n- Do not run full dataset downloads or raw-data submodule cloning unless the\n  user has accepted network and storage costs. Use tiny fixtures and the\n  bundled validator for bounded checks.\n\n## Validation expectations\n\n- Builder `--help` commands are safe and should list the documented options.\n- Full Sudoku and Maze conversion require network access to Hugging Face\ndatasets.\n- Full ARC conversion requires the ARC\u002FConceptARC raw-data submodules or an\n  equivalent local directory.\n- Converted train\u002Ftest split directories must contain `dataset.json` plus\n  `\u003Csubset>__inputs.npy`, `\u003Csubset>__labels.npy`,\n  `\u003Csubset>__puzzle_identifiers.npy`, `\u003Csubset>__puzzle_indices.npy`, and\n  `\u003Csubset>__group_indices.npy` for each subset in metadata `sets`.\n","# HRM Data Preparation\n\nUse this sub-skill when the task is to create, inspect, validate, or visualize\nHierarchical Reasoning Model (HRM) datasets for ARC, Sudoku, or Maze workflows.\nThe original repo builders produce a shared directory layout consumed by\n`PuzzleDataset` and by `pretrain.py`\u002F`evaluate.py`.\n\n## When to use\n\n- The user asks to build ARC-AGI, ARC-AGI-2, ConceptARC, Sudoku Extreme, or Maze\n  datasets for HRM.\n- The task names `dataset\u002Fbuild_arc_dataset.py`, `build_sudoku_dataset.py`,\n  `build_maze_dataset.py`, `PuzzleDataset`, `PuzzleDatasetMetadata`,\n  `identifiers.json`, or `all__inputs.npy` style files.\n- A training or evaluation run fails before model execution because dataset\n  paths, metadata, `.npy` arrays, split\u002Fsubset names, token ranges, or puzzle\n  index arrays are wrong.\n- The user wants a browser visualizer for converted puzzle folders.\n\n## Route map\n\n1. Read [references\u002Fdata-formats.md](references\u002Fdata-formats.md) for the exact\n   shared dataset directory structure, metadata fields, array meanings, and\n   validation invariants.\n2. Read [references\u002Fworkflows.md](references\u002Fworkflows.md) for ARC, Sudoku,\n   Maze, and visualization command recipes, including network\u002Fsubmodule\n   prerequisites and safe bounded variants.\n3. Use [references\u002Ftroubleshooting.md](references\u002Ftroubleshooting.md) when a\n   builder cannot find raw data, Hugging Face downloads fail, token ranges look\n   invalid, or `PuzzleDataset` yields no batches.\n4. Run [scripts\u002Fvalidate_dataset_layout.py](scripts\u002Fvalidate_dataset_layout.py)\n   against a converted dataset before long training or evaluation.\n5. Use [scripts\u002Frun_dataset_builder.py](scripts\u002Frun_dataset_builder.py) when a\n   future task needs a wrapper that forwards arguments to an HRM checkout from\n   any current working directory.\n6. Open [scripts\u002Fpuzzle_visualizer.html](scripts\u002Fpuzzle_visualizer.html) in a\n   browser for local inspection of generated puzzle folders. The required\n   `npyjs` asset is bundled under `scripts\u002Fassets\u002F`.\n\n## Core workflow\n\n```bash\n# ARC-1 from initialized ARC-AGI + ConceptARC submodules\npython dataset\u002Fbuild_arc_dataset.py\n\n# ARC-2 from ARC-AGI-2 raw data\npython dataset\u002Fbuild_arc_dataset.py \\\n  --dataset-dirs dataset\u002Fraw-data\u002FARC-AGI-2\u002Fdata \\\n  --output-dir data\u002Farc-2-aug-1000\n\n# Sudoku 1K augmented demo\npython dataset\u002Fbuild_sudoku_dataset.py \\\n  --output-dir data\u002Fsudoku-extreme-1k-aug-1000 \\\n  --subsample-size 1000 \\\n  --num-aug 1000\n\n# Maze 30x30 hard 1K\npython dataset\u002Fbuild_maze_dataset.py --output-dir data\u002Fmaze-30x30-hard-1k\n```\n\nAfter building, validate the shared layout before training:\n\n```bash\npython \u003Cskill>\u002Fsub-skills\u002Fdata-preparation\u002Fscripts\u002Fvalidate_dataset_layout.py \\\n  data\u002Fsudoku-extreme-1k-aug-1000 --splits train test\n```\n\nReplace `\u003Cskill>` with the root of this generated skill directory. The helper is\nself-contained; it does not need the original HRM checkout unless you are\nrunning the original builder scripts.\n\n## Boundaries\n\n- This sub-skill owns dataset builders, converted layout, schema validation,\n  and local visualization.\n- Use `model-architecture` for HRM ACT v1 config fields, losses, model class\n  routing, FlashAttention layers, and sparse puzzle embeddings.\n- Use `training-evaluation` for `pretrain.py`, `evaluate.py`, checkpoints,\n  W&B, CUDA training, and ARC prediction post-processing.\n- Do not run full dataset downloads or raw-data submodule cloning unless the\n  user has accepted network and storage costs. Use tiny fixtures and the\n  bundled validator for bounded checks.\n\n## Validation expectations\n\n- Builder `--help` commands are safe and should list the documented options.\n- Full Sudoku and Maze conversion require network access to Hugging Face\ndatasets.\n- Full ARC conversion requires the ARC\u002FConceptARC raw-data submodules or an\n  equivalent local directory.\n- Converted train\u002Ftest split directories must contain `dataset.json` plus\n  `\u003Csubset>__inputs.npy`, `\u003Csubset>__labels.npy`,\n  `\u003Csubset>__puzzle_identifiers.npy`, `\u003Csubset>__puzzle_indices.npy`, and\n  `\u003Csubset>__group_indices.npy` for each subset in metadata `sets`.\n",{"name":5,"license":24,"metadata":25,"description":7,"disable-model-invocation":27},"Apache 2.0",{"disco-role":26},"operating",true,"main","d69096e239f2a89aabc2cd078c47664a9d3b1e11","06-前端与设计\u002Fhrm\u002Fsub-skills\u002Fdata-preparation",3,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts","https:\u002F\u002Fgithub.com\u002Flza6","个人vx：Tf00798\r\n个人✈：@TFGY999","2026-10-08 14:29:22",{"source":8,"skillName":5,"commands":37,"prompt":47,"zipAvailable":27},[38,41,44],{"tool":39,"command":40},"npx","npx skills add lza6\u002FArmor-piercing-system-prompts --skill data-preparation",{"tool":42,"command":43},"bunx","bunx skills add lza6\u002FArmor-piercing-system-prompts --skill data-preparation",{"tool":45,"command":46},"pnpm","pnpm dlx skills add lza6\u002FArmor-piercing-system-prompts --skill data-preparation","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F06-前端与设计\u002Fhrm\u002Fsub-skills\u002Fdata-preparation\u002FSKILL.md",[49,55,59],{"path":50,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":53,"sourceHtmlUrl":54},"references\u002Fdata-formats.md","REFERENCE","参考文档",4031,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F06-前端与设计\u002Fhrm\u002Fsub-skills\u002Fdata-preparation\u002Freferences\u002Fdata-formats.md",{"path":56,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":57,"sourceHtmlUrl":58},"references\u002Ftroubleshooting.md",3262,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F06-前端与设计\u002Fhrm\u002Fsub-skills\u002Fdata-preparation\u002Freferences\u002Ftroubleshooting.md",{"path":60,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":61,"sourceHtmlUrl":62},"references\u002Fworkflows.md",3298,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F06-前端与设计\u002Fhrm\u002Fsub-skills\u002Fdata-preparation\u002Freferences\u002Fworkflows.md",[64,72,77,84,91,98],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":68,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M7FFG485KY83TD6HSC4YPCMC","obsidian-clipper-template-creator","Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.","MIT","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobsidian-clipper-template-creator\u002FSKILL.md",[],[],{"skillNo":73,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":68,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M7FFG0T6AZ0STEJPSB3CX84K","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobsidian-clipper-template-creator-antigravity-awesome-skills-main\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":68,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":19},"01M7FFFY0QEM14SKWPDYGMXM2P","python-api","Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP\u002FmAP semantics, and source-style import diagnostics.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobject-detection-metrics\u002Fsub-skills\u002Fpython-api\u002FSKILL.md",[],[],{"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":68,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":19},"01M7FFFVZQZ2GJ0C0TGP7JHD20","file-evaluation","Folder-based PASCAL VOC AP\u002FmAP evaluation from ground-truth and detection text files.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobject-detection-metrics\u002Fsub-skills\u002Ffile-evaluation\u002FSKILL.md",[],[],{"skillNo":92,"name":93,"nameZh":6,"description":94,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":68,"licenseFileUrl":6,"sourceHtmlUrl":95,"categories":96,"keywords":97,"publishedAt":19},"01M7FFFSFNYSE77SJ6PE1CK2DA","object-detection-metrics","Guides legacy Object-Detection-Metrics PASCAL VOC AP\u002FmAP evaluation from detection text folders and direct Python API classes.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobject-detection-metrics\u002FSKILL.md",[],[],{"skillNo":99,"name":100,"nameZh":6,"description":101,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":68,"licenseFileUrl":6,"sourceHtmlUrl":102,"categories":103,"keywords":104,"publishedAt":19},"01M7FFFP0GCK1RVXBCPQ8APBMT","nx-workspace-patterns","Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fnx-workspace-patterns\u002FSKILL.md",[],[]]