[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7C2AMR7E2JTHQFGR5ZN2BHE":3,"skill-same:01M7C2AMR7E2JTHQFGR5ZN2BHE":45},{"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":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":30,"install":31,"resources":44},"01M7C2AMR7E2JTHQFGR5ZN2BHE","output-validation",null,"Local self-check of instructions and mask outputs (format\u002Frange\u002Fconsistency) without using GT.","linyouxue\u002Fskill-repair-benchmark","linyouxue","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F216151464?v=4","01M7BJEVCXKD74KH6FG9CBVSQB","林宥学","01M7BJEVE02NP217V93BKZFC0J",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Flinyouxue\u002Fskill-repair-benchmark\u002Fblob\u002Fmain\u002Fevaluation\u002Fresults\u002FWML-Opus47-15Tasks-20260923\u002Ftasks\u002Fdynamic-object-aware-egomotion\u002Frepaired_skill\u002Foutput-validation\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: output-validation\ndescription: \"Local self-check of instructions and mask outputs (format\u002Frange\u002Fconsistency) without using GT.\"\n---\n\n# When to use\n- After generating your outputs (interval instructions, masks, etc.), before submission\u002Fhand-off.\n\n# Checks\n- **Key format**: every key is `\"{start}->{end}\"`, integers only, `start\u003C=end`.\n- **Coverage**: max frame index ≤ video total-1; consistent with your sampling policy.\n- **Frame count**: NPZ `f_{i}_*` count equals sampled frame count; no gaps or missing components.\n- **CSR integrity**: each frame has `data\u002Findices\u002Findptr`; `len(indptr)==H+1`; `indptr[-1]==indices.size`; indices within `[0,W)`.\n- **Value validity**: JSON values are non-empty string lists; labels in the allowed set.\n\n## Payload plausibility (additive; not a correctness proof)\nFormat checks alone will accept degenerate payloads (all-empty masks, all-full masks, wildly inconsistent frame-to-frame outputs). For per-frame mask\u002Flabel artifacts, also compute cheap self-consistency statistics and use them as *retry triggers*, not as ground-truth assertions:\n- **Foreground-fraction distribution**: per-frame nnz \u002F (H·W). Flag if the mean falls outside a documented band for the task, or if any individual frame is ~0% or near-saturated (e.g., ≥ ~40%) when neither extreme is expected.\n- **Empty-frame ratio**: fraction of frames with nnz == 0. Flag when a non-trivial share of frames is empty in a scene where motion labels other than `Stay` are present.\n- **Temporal flicker**: mean of `logical_xor(mask_i, mask_{i-1}).mean()` across consecutive sampled frames. Flag unusually high values (masks that flip most of their pixels every frame).\n- **Cross-artifact consistency**: if the motion-label stream contains non-`Stay` labels for a span but the corresponding mask frames are all empty (or vice versa), flag the mismatch.\n\nIf any plausibility signal fires, do **not** ship. Route back to the method-selection\u002Ftuning step in the producing skill (e.g., `dyn-object-masks` *Method selection & tuning order*) and re-run.\n\n### Scope and non-goals\n- Plausibility bands are self-consistency triggers for retry; they are *not* equality assertions against unknown ground truth and must not be turned into hard pass\u002Ffail rules pretending to score correctness.\n- Do not modify or weaken the schema\u002Fformat checks above; plausibility checks are strictly additive.\n\n# Reference snippet\n```python\nimport json, numpy as np, cv2\nVIDEO_PATH = \"\u003Cpath\u002Fto\u002Fvideo>\"\nINSTRUCTIONS_PATH = \"\u003Cpath\u002Fto\u002Finterval_instructions.json>\"\nMASKS_PATH = \"\u003Cpath\u002Fto\u002Fmasks.npz>\"\ncap=cv2.VideoCapture(VIDEO_PATH)\nn=int(cap.get(cv2.CAP_PROP_FRAME_COUNT)); H=int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT)); W=int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))\nj=json.load(open(INSTRUCTIONS_PATH))\nnpz=np.load(MASKS_PATH)\nfor k,v in j.items():\n    s,e=k.split(\"->\"); assert s.isdigit() and e.isdigit()\n    s=int(s); e=int(e); assert 0\u003C=s\u003C=e\u003Cn\n    for lbl in v: assert isinstance(lbl,str)\nframes=0\nwhile f\"f_{frames}_data\" in npz: frames+=1\nassert frames>0\nassert npz[\"shape\"][0]==H and npz[\"shape\"][1]==W\nindptr=npz[\"f_0_indptr\"]; indices=npz[\"f_0_indices\"]\nassert indptr.shape[0]==H+1 and indptr[-1]==indices.size\nassert indices.size==0 or (indices.min()>=0 and indices.max()\u003CW)\n\n# Payload plausibility (retry triggers, not correctness assertions)\nHW = float(H*W)\nfracs, prev = [], None\nflicker = []\nfor i in range(frames):\n    ip = npz[f\"f_{i}_indptr\"]; idx = npz[f\"f_{i}_indices\"]\n    nnz = int(ip[-1])\n    fracs.append(nnz \u002F HW)\n    # dense reconstruction only for flicker; skip if memory-bound\n    m = np.zeros((H, W), dtype=bool)\n    for r in range(H):\n        a, b = ip[r], ip[r+1]\n        if b > a: m[r, idx[a:b]] = True\n    if prev is not None:\n        flicker.append(float(np.logical_xor(m, prev).mean()))\n    prev = m\nempty_ratio = sum(1 for f in fracs if f == 0) \u002F max(1, len(fracs))\nmean_frac = float(np.mean(fracs)) if fracs else 0.0\nmean_flicker = float(np.mean(flicker)) if flicker else 0.0\n# Compare mean_frac, empty_ratio, mean_flicker against the task's documented band and\n# trigger a retry (not an assertion failure) if any is out of range.\n```\n\n# Self-check list\n- [ ] JSON keys\u002Fvalues pass format checks.\n- [ ] Max frame index within video range and near sampled max.\n- [ ] NPZ frame count matches sampling; keys consecutive.\n- [ ] CSR structure and `shape` validated.\n- [ ] Payload plausibility computed: foreground-fraction distribution, empty-frame ratio, and temporal flicker are all inside the documented band, and mask activity is consistent with the motion-label stream. If not, route back to the producing skill and retry before submission.\n","# When to use\n- After generating your outputs (interval instructions, masks, etc.), before submission\u002Fhand-off.\n\n# Checks\n- **Key format**: every key is `\"{start}->{end}\"`, integers only, `start\u003C=end`.\n- **Coverage**: max frame index ≤ video total-1; consistent with your sampling policy.\n- **Frame count**: NPZ `f_{i}_*` count equals sampled frame count; no gaps or missing components.\n- **CSR integrity**: each frame has `data\u002Findices\u002Findptr`; `len(indptr)==H+1`; `indptr[-1]==indices.size`; indices within `[0,W)`.\n- **Value validity**: JSON values are non-empty string lists; labels in the allowed set.\n\n## Payload plausibility (additive; not a correctness proof)\nFormat checks alone will accept degenerate payloads (all-empty masks, all-full masks, wildly inconsistent frame-to-frame outputs). For per-frame mask\u002Flabel artifacts, also compute cheap self-consistency statistics and use them as *retry triggers*, not as ground-truth assertions:\n- **Foreground-fraction distribution**: per-frame nnz \u002F (H·W). Flag if the mean falls outside a documented band for the task, or if any individual frame is ~0% or near-saturated (e.g., ≥ ~40%) when neither extreme is expected.\n- **Empty-frame ratio**: fraction of frames with nnz == 0. Flag when a non-trivial share of frames is empty in a scene where motion labels other than `Stay` are present.\n- **Temporal flicker**: mean of `logical_xor(mask_i, mask_{i-1}).mean()` across consecutive sampled frames. Flag unusually high values (masks that flip most of their pixels every frame).\n- **Cross-artifact consistency**: if the motion-label stream contains non-`Stay` labels for a span but the corresponding mask frames are all empty (or vice versa), flag the mismatch.\n\nIf any plausibility signal fires, do **not** ship. Route back to the method-selection\u002Ftuning step in the producing skill (e.g., `dyn-object-masks` *Method selection & tuning order*) and re-run.\n\n### Scope and non-goals\n- Plausibility bands are self-consistency triggers for retry; they are *not* equality assertions against unknown ground truth and must not be turned into hard pass\u002Ffail rules pretending to score correctness.\n- Do not modify or weaken the schema\u002Fformat checks above; plausibility checks are strictly additive.\n\n# Reference snippet\n```python\nimport json, numpy as np, cv2\nVIDEO_PATH = \"\u003Cpath\u002Fto\u002Fvideo>\"\nINSTRUCTIONS_PATH = \"\u003Cpath\u002Fto\u002Finterval_instructions.json>\"\nMASKS_PATH = \"\u003Cpath\u002Fto\u002Fmasks.npz>\"\ncap=cv2.VideoCapture(VIDEO_PATH)\nn=int(cap.get(cv2.CAP_PROP_FRAME_COUNT)); H=int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT)); W=int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))\nj=json.load(open(INSTRUCTIONS_PATH))\nnpz=np.load(MASKS_PATH)\nfor k,v in j.items():\n    s,e=k.split(\"->\"); assert s.isdigit() and e.isdigit()\n    s=int(s); e=int(e); assert 0\u003C=s\u003C=e\u003Cn\n    for lbl in v: assert isinstance(lbl,str)\nframes=0\nwhile f\"f_{frames}_data\" in npz: frames+=1\nassert frames>0\nassert npz[\"shape\"][0]==H and npz[\"shape\"][1]==W\nindptr=npz[\"f_0_indptr\"]; indices=npz[\"f_0_indices\"]\nassert indptr.shape[0]==H+1 and indptr[-1]==indices.size\nassert indices.size==0 or (indices.min()>=0 and indices.max()\u003CW)\n\n# Payload plausibility (retry triggers, not correctness assertions)\nHW = float(H*W)\nfracs, prev = [], None\nflicker = []\nfor i in range(frames):\n    ip = npz[f\"f_{i}_indptr\"]; idx = npz[f\"f_{i}_indices\"]\n    nnz = int(ip[-1])\n    fracs.append(nnz \u002F HW)\n    # dense reconstruction only for flicker; skip if memory-bound\n    m = np.zeros((H, W), dtype=bool)\n    for r in range(H):\n        a, b = ip[r], ip[r+1]\n        if b > a: m[r, idx[a:b]] = True\n    if prev is not None:\n        flicker.append(float(np.logical_xor(m, prev).mean()))\n    prev = m\nempty_ratio = sum(1 for f in fracs if f == 0) \u002F max(1, len(fracs))\nmean_frac = float(np.mean(fracs)) if fracs else 0.0\nmean_flicker = float(np.mean(flicker)) if flicker else 0.0\n# Compare mean_frac, empty_ratio, mean_flicker against the task's documented band and\n# trigger a retry (not an assertion failure) if any is out of range.\n```\n\n# Self-check list\n- [ ] JSON keys\u002Fvalues pass format checks.\n- [ ] Max frame index within video range and near sampled max.\n- [ ] NPZ frame count matches sampling; keys consecutive.\n- [ ] CSR structure and `shape` validated.\n- [ ] Payload plausibility computed: foreground-fraction distribution, empty-frame ratio, and temporal flicker are all inside the documented band, and mask activity is consistent with the motion-label stream. If not, route back to the producing skill and retry before submission.\n",{"name":5,"description":7},"main","cdcb6d230f3b15c3c33f8a2bdd16439f43602cbe","evaluation\u002Fresults\u002FWML-Opus47-15Tasks-20260923\u002Ftasks\u002Fdynamic-object-aware-egomotion\u002Frepaired_skill\u002Foutput-validation",1,"https:\u002F\u002Fgithub.com\u002Flinyouxue\u002Fskill-repair-benchmark","https:\u002F\u002Fgithub.com\u002Flinyouxue","2026-11-14 20:28:35",{"source":8,"skillName":5,"commands":32,"prompt":42,"zipAvailable":43},[33,36,39],{"tool":34,"command":35},"npx","npx skills add linyouxue\u002Fskill-repair-benchmark --skill output-validation",{"tool":37,"command":38},"bunx","bunx skills add linyouxue\u002Fskill-repair-benchmark --skill output-validation",{"tool":40,"command":41},"pnpm","pnpm dlx skills add linyouxue\u002Fskill-repair-benchmark --skill output-validation","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Flinyouxue\u002Fskill-repair-benchmark\u002Fblob\u002Fmain\u002Fevaluation\u002Fresults\u002FWML-Opus47-15Tasks-20260923\u002Ftasks\u002Fdynamic-object-aware-egomotion\u002Frepaired_skill\u002Foutput-validation\u002FSKILL.md",true,[],[46,53,60,67,72,77],{"skillNo":47,"name":48,"nameZh":6,"description":49,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":50,"categories":51,"keywords":52,"publishedAt":19},"01M7E1HGVRY73NBJ022CJG13X6","video-frame-extraction","Extract frames from video files and save them as images using OpenCV","https:\u002F\u002Fgithub.com\u002Flinyouxue\u002Fskill-repair-benchmark\u002Fblob\u002Fmain\u002Fjpg-ocr-stat\u002Fresult\u002Fskilltta-claude4.7\u002Fgroup_03\u002Frun_03\u002Finputs\u002Fskills\u002Fvideo-frame-extraction\u002FSKILL.md",[],[],{"skillNo":54,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":57,"categories":58,"keywords":59,"publishedAt":19},"01M7E1HAG5R7CJ9C8NYJS771MB","openai-vision","Analyze images and multi-frame sequences using OpenAI GPT vision models","https:\u002F\u002Fgithub.com\u002Flinyouxue\u002Fskill-repair-benchmark\u002Fblob\u002Fmain\u002Fjpg-ocr-stat\u002Fresult\u002Fskilltta-claude4.7\u002Fgroup_03\u002Frun_03\u002Finputs\u002Fskills\u002Fopenai-vision\u002FSKILL.md",[],[],{"skillNo":61,"name":62,"nameZh":6,"description":63,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":64,"categories":65,"keywords":66,"publishedAt":19},"01M7E1H7DADKB6QBWKC2KQ63SZ","image-ocr","Extract text content from images using Tesseract OCR via Python, including structured field extraction from receipts and forms","https:\u002F\u002Fgithub.com\u002Flinyouxue\u002Fskill-repair-benchmark\u002Fblob\u002Fmain\u002Fjpg-ocr-stat\u002Fresult\u002Fskilltta-claude4.7\u002Fgroup_03\u002Frun_03\u002Finputs\u002Fskills\u002Fimage-ocr\u002FSKILL.md",[],[],{"skillNo":68,"name":48,"nameZh":6,"description":49,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M7E1H0WBNT43TENGQY7AVZHD","https:\u002F\u002Fgithub.com\u002Flinyouxue\u002Fskill-repair-benchmark\u002Fblob\u002Fmain\u002Fjpg-ocr-stat\u002Fresult\u002Fskilltta-claude4.7\u002Fgroup_03\u002Frun_02\u002Finputs\u002Fskills\u002Fvideo-frame-extraction\u002FSKILL.md",[],[],{"skillNo":73,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M7E1GTC7D4N4Q2EF03875J9E","https:\u002F\u002Fgithub.com\u002Flinyouxue\u002Fskill-repair-benchmark\u002Fblob\u002Fmain\u002Fjpg-ocr-stat\u002Fresult\u002Fskilltta-claude4.7\u002Fgroup_03\u002Frun_02\u002Finputs\u002Fskills\u002Fopenai-vision\u002FSKILL.md",[],[],{"skillNo":78,"name":62,"nameZh":6,"description":63,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":19},"01M7E1GQ92CWFMGH53KSSE4WER","https:\u002F\u002Fgithub.com\u002Flinyouxue\u002Fskill-repair-benchmark\u002Fblob\u002Fmain\u002Fjpg-ocr-stat\u002Fresult\u002Fskilltta-claude4.7\u002Fgroup_03\u002Frun_02\u002Finputs\u002Fskills\u002Fimage-ocr\u002FSKILL.md",[],[]]