[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7FDRXZ2CMXZX1DMNSWAJ0QZ":3,"skill-same:01M7FDRXZ2CMXZX1DMNSWAJ0QZ":46},{"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":30,"updatedAt":31,"install":32,"resources":45},"01M7FDRXZ2CMXZX1DMNSWAJ0QZ","orx-lit-review",null,"Search and read research papers. The main agent calls alphaXiv, OpenAlex, and bioRxiv discovery primitives, ranks the combined candidates, and chooses sources for focused follow-ups. Use for literature reviews, related work, prior art, papers, authors, methods, benchmarks, or research claims; never delegate the retrieval loop to a sub-agent.","lza6\u002FArmor-piercing-system-prompts","lza6","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F128880206?v=4","01M7EB3K7WPX5T28EJ2NVHE6EG","Chinese-tingfeng","01M7EB3K97E1KQ16QWKJMFNGX8",11,"MIT","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002FalphaXiv__OpenResearch\u002Fagent-skills\u002Forx-lit-review\u002FSKILL.md",[],[],"2026-11-16 10:05:34",false,"---\nname: orx-lit-review\ndescription: \"Search and read research papers. The main agent calls alphaXiv, OpenAlex, and bioRxiv discovery primitives, ranks the combined candidates, and chooses sources for focused follow-ups. Use for literature reviews, related work, prior art, papers, authors, methods, benchmarks, or research claims; never delegate the retrieval loop to a sub-agent.\"\n---\n\n# Literature retrieval\n\nYou are the retrieval ranker. Call the alphaXiv, OpenAlex, and bioRxiv\nprimitives yourself, inspect the returned candidates, and decide which sources\nare useful for each focused follow-up. Never delegate this loop to a sub-agent.\n\nEach command performs exactly one public endpoint request and emits its\nstructured JSON result. No login is required:\n\n```sh\norx discover keyword \"\u003Cexact keyword query>\"\norx discover embedding \"\u003Csemantic description in the user's terms>\"\norx discover openalex \"\u003Cscholarly search query>\"\norx discover biorxiv \"\u003Cbiology preprint query>\"\n```\n\n- `keyword` searches title, abstract, and full text. Results include the match\n  snippets that explain why each paper was retrieved. Use short exact terms:\n  method names, acronyms, benchmarks, authors, or title phrases. Use only terms\n  stated by the user or observed in results; never invent an acronym expansion.\n- `embedding` searches titles and abstracts semantically, then reranks by\n  similarity and the requested priority. Use the user's actual question or a\n  concise description of a genuinely missing facet.\n- `openalex` searches the cross-disciplinary OpenAlex scholarly graph. It is\n  especially useful for journal\u002Fconference papers, citation context, and work\n  outside arXiv.\n- `biorxiv` searches OpenAlex's bioRxiv source index. bioRxiv has no comparable\n  native search API; the bioRxiv API is used later when reading a selected DOI.\n- Every primitive returns the same JSON shape: `source`, self-routing `id`,\n  title, abstract, and publication date. alphaXiv results may include votes and\n  full-text snippets; OpenAlex and bioRxiv results may include citations.\n\n## Date and ranking controls\n\nRetrieval is not date-bounded unless you supply a bound. Add the same controls\nto any primitive when the question calls for them:\n\n```sh\norx discover keyword \"\u003Cquery>\" --published-after 2024-01-01 --prioritize recency\norx discover embedding \"\u003Cquery>\" --published-before 2012-01-01 --prioritize historical\norx discover openalex \"\u003Cquery>\" --published-after 2024-01-01 --prioritize recency\norx discover biorxiv \"\u003Cquery>\" --limit 20\n```\n\n- `--published-after` and `--published-before` are inclusive `YYYY-MM-DD`\n  bounds. Do not invent a cutoff merely to favour newer work.\n- Older or narrow `--published-before` embedding searches can return a thin or\n  empty candidate set because the upper bound is applied after vector retrieval.\n  Report what comes back; do not treat an empty set as proof that no literature\n  exists or retry the identical query and window.\n- `--prioritize` is `default`, `recency`, `historical`, or `popular`.\n- `--limit` can narrow alphaXiv output but cannot widen alphaXiv's fixed\n  server-side candidate pools. For OpenAlex and bioRxiv it also controls the\n  requested pool size.\n- OpenAlex and bioRxiv implement these controls with OpenAlex publication-date\n  filters, then rerank the returned relevance pool by date or citations. Their\n  ranking is best-effort and is not identical to alphaXiv's semantic,\n  vote-aware ranking.\n- Use `recency` for explicitly new\u002Flatest work. Use `historical` for seminal or\n  foundational work. Use `popular` only when the user asks about votes,\n  popularity, or community standing.\n\n## Main-agent retrieval loop\n\nYou are the low-latency retrieval ranker. Run the loop below yourself.\n\n### Set up the retrieval query\n\n1. If using keyword retrieval, build focused terms using only wording from the\n   user or prior tool results. Never guess an acronym expansion. General-purpose\n   padding reduces result quality.\n2. For semantic or scholarly-graph retrieval, build one short faithful question\n   in the user's terms rather than a padded reformulation.\n3. Estimate retrieval difficulty from 1–10. This controls a budget of complete\n   follow-up rounds: difficulty 1–3 gets 0 rounds, 4–7 gets 1, and 8–10 gets 2.\n4. Resolve one publication window and priority for the request. Every initial\n   and follow-up call must inherit those exact controls; never widen a window or\n   change priority during the loop. Every returned candidate already satisfies\n   that window, so rank what is available instead of lamenting well-known work\n   that the user excluded.\n\n### Run and rank\n\n1. Choose the initial sources and strategies that fit the query. For arXiv-heavy\n   ML, CS, math, or physics questions, use alphaXiv keyword, embedding, or both\n   according to whether exact full-text evidence, semantic coverage, or both are\n   useful. Add OpenAlex for broader journal, conference, citation, or\n   cross-disciplinary coverage. Use bioRxiv for biology and adjacent\n   life-science preprints, not as a ritual call for unrelated topics. When the\n   corpus is genuinely ambiguous or interdisciplinary, query multiple relevant\n   sources concurrently. If the initial round includes alphaXiv keyword and its\n   terms mix other terms with one or more 2–10 character tokens\n   that start with a letter, contain only letters, digits, or hyphens, and have\n   at least two uppercase letters, concurrently run one additional keyword call\n   whose query is exactly those acronym tokens joined by spaces and nothing\n   else. This recovery call is part of the initial round.\n2. Treat initial calls independently: retain every successful result set when\n   another call fails. If none returns results and follow-up budget remains,\n   use a round only when a focused recovery query is likely to work.\n3. Inspect and deduplicate every candidate. Match exact `id` first, then a DOI\n   or arXiv id visible in the metadata, then exact normalized title as a\n   cross-source fallback. Prefer the alphaXiv representation of an arXiv\n   duplicate because it supports full-text reading. bioRxiv is a subset of\n   OpenAlex, so overlap between those calls is expected. Within each source,\n   the API order already blends topical relevance with the requested priority:\n   - With `recency`, freshness is already upranked and old accumulated votes\n     are damped. Reorder only for topical fit; do not exclude an older but much\n     better match.\n   - With `popular`, votes or citations dominate among topically plausible\n     results. Keep high-impact relevant papers, but drop off-topic ones.\n   - Otherwise, topical relevance remains primary with freshness and votes\n     already nudging the order. Do not apply those preferences a second time.\n4. If the initial candidates provide solid topical coverage, stop immediately\n   and rank 5–15 IDs. Fast and slightly less complete is better than an\n   exploratory search. Prefer fewer strong papers over padding.\n5. Otherwise, spend at most the difficulty-derived number of follow-up rounds.\n   One round targets one concrete missing acronym, method, benchmark,\n   organization, title phrase, venue, or subtopic. Choose one or more sources\n   based on the gap: alphaXiv keyword for exact\u002Ffull-text evidence, alphaXiv\n   embedding for a semantic arXiv angle, OpenAlex for broad scholarly or\n   citation coverage, and bioRxiv for recent biology preprints. Later rounds do\n   **not** need to query all sources. Calls for the same missing angle count\n   together as one round. Never spend a round merely rephrasing an existing\n   search. Re-evaluate after each round and stop as soon as coverage is\n   sufficient. The budget is a hard cap, not a target.\n6. Drop each selected ID that did not appear in a successful initial or\n   follow-up result, retaining the surviving IDs in your chosen rank order. If\n   no selected ID survives, fall back to the first 15 unique IDs in observation\n   order, with initial results before follow-up results. Never invent or recall\n   an ID.\n\nBatch all facets into one broad retrieval loop and plan against a cap of two\ncomplete loops per user turn. If a genuinely distinct topic still forces a\nthird or fourth loop, run it in shallow mode: initial searches only, with zero\nfollow-up rounds. This degradation is a backstop, not permission to plan extra\nloops. Refuse a fifth loop and answer from the papers already found.\n\nFor a set-of-papers request such as “find papers,” “top papers,” “what is out\nthere,” or “what should I read,” return the ranked discovery results and stop.\nDepth on individual papers is not part of the discovery loop. When the request\ninstead needs claim-level synthesis, methodological details, or comparison,\nfinish retrieval first and then read the 3–5 most load-bearing candidates with\n`orx paper \u003Cid>` (or the number the user requested). Do not narrow to 3–5\npapers before retrieval has produced its ranked 5–15 candidate set.\n\nDo not compare alphaXiv votes numerically with OpenAlex citations; they measure\ndifferent things. Topical fit is the cross-source ranking signal.\n\nIn the final answer, link every alphaXiv\u002FarXiv paper title or paper ID to\n`https:\u002F\u002Fwww.alphaxiv.org\u002Fabs\u002F\u003Cversionless-paperId>`. Never return an\n`arxiv.org` link for those papers. Link a DOI result to `https:\u002F\u002Fdoi.org\u002F\u003Cdoi>`\nand a bare OpenAlex `W…` id to `https:\u002F\u002Fopenalex.org\u002F\u003Cid>`.\n\nFor claim-level synthesis, place the supporting source link immediately after\neach substantive scholarly claim, and use a paper as claim-level support only\nafter reading it. A discovery-only result list may link candidate titles, but\nmust not imply that their methods or findings were verified from snippets alone.\n\n## Reading selected papers\n\n`orx paper` auto-detects an arXiv id\u002FURL, bioRxiv DOI, other DOI, or OpenAlex\n`W…` id. For alphaXiv it returns a compact structured report; use `--full` only\nwhen you explicitly need raw text even if a report exists. Without `--full`, a\nmissing report automatically falls back to extracted full text in the same\ncommand. `--full` skips the report entirely rather than acting as a superset of\nthe default. If extracted text is also unavailable, use the alphaXiv paper link\nit returns.\n\n`orx paper` prints the alphaXiv link before the content. When alphaXiv has an\nassociated repository, it then prints `GitHub: \u003Curl>`. This is the most-starred\nassociated repository and can be a framework rather than the paper's own code,\nso sanity-check it before treating it as the implementation.\n\nAll discovery and paper commands honor the user's disabled literature-source\nsettings; do not work around an error saying a source is disabled.\n","# Literature retrieval\n\nYou are the retrieval ranker. Call the alphaXiv, OpenAlex, and bioRxiv\nprimitives yourself, inspect the returned candidates, and decide which sources\nare useful for each focused follow-up. Never delegate this loop to a sub-agent.\n\nEach command performs exactly one public endpoint request and emits its\nstructured JSON result. No login is required:\n\n```sh\norx discover keyword \"\u003Cexact keyword query>\"\norx discover embedding \"\u003Csemantic description in the user's terms>\"\norx discover openalex \"\u003Cscholarly search query>\"\norx discover biorxiv \"\u003Cbiology preprint query>\"\n```\n\n- `keyword` searches title, abstract, and full text. Results include the match\n  snippets that explain why each paper was retrieved. Use short exact terms:\n  method names, acronyms, benchmarks, authors, or title phrases. Use only terms\n  stated by the user or observed in results; never invent an acronym expansion.\n- `embedding` searches titles and abstracts semantically, then reranks by\n  similarity and the requested priority. Use the user's actual question or a\n  concise description of a genuinely missing facet.\n- `openalex` searches the cross-disciplinary OpenAlex scholarly graph. It is\n  especially useful for journal\u002Fconference papers, citation context, and work\n  outside arXiv.\n- `biorxiv` searches OpenAlex's bioRxiv source index. bioRxiv has no comparable\n  native search API; the bioRxiv API is used later when reading a selected DOI.\n- Every primitive returns the same JSON shape: `source`, self-routing `id`,\n  title, abstract, and publication date. alphaXiv results may include votes and\n  full-text snippets; OpenAlex and bioRxiv results may include citations.\n\n## Date and ranking controls\n\nRetrieval is not date-bounded unless you supply a bound. Add the same controls\nto any primitive when the question calls for them:\n\n```sh\norx discover keyword \"\u003Cquery>\" --published-after 2024-01-01 --prioritize recency\norx discover embedding \"\u003Cquery>\" --published-before 2012-01-01 --prioritize historical\norx discover openalex \"\u003Cquery>\" --published-after 2024-01-01 --prioritize recency\norx discover biorxiv \"\u003Cquery>\" --limit 20\n```\n\n- `--published-after` and `--published-before` are inclusive `YYYY-MM-DD`\n  bounds. Do not invent a cutoff merely to favour newer work.\n- Older or narrow `--published-before` embedding searches can return a thin or\n  empty candidate set because the upper bound is applied after vector retrieval.\n  Report what comes back; do not treat an empty set as proof that no literature\n  exists or retry the identical query and window.\n- `--prioritize` is `default`, `recency`, `historical`, or `popular`.\n- `--limit` can narrow alphaXiv output but cannot widen alphaXiv's fixed\n  server-side candidate pools. For OpenAlex and bioRxiv it also controls the\n  requested pool size.\n- OpenAlex and bioRxiv implement these controls with OpenAlex publication-date\n  filters, then rerank the returned relevance pool by date or citations. Their\n  ranking is best-effort and is not identical to alphaXiv's semantic,\n  vote-aware ranking.\n- Use `recency` for explicitly new\u002Flatest work. Use `historical` for seminal or\n  foundational work. Use `popular` only when the user asks about votes,\n  popularity, or community standing.\n\n## Main-agent retrieval loop\n\nYou are the low-latency retrieval ranker. Run the loop below yourself.\n\n### Set up the retrieval query\n\n1. If using keyword retrieval, build focused terms using only wording from the\n   user or prior tool results. Never guess an acronym expansion. General-purpose\n   padding reduces result quality.\n2. For semantic or scholarly-graph retrieval, build one short faithful question\n   in the user's terms rather than a padded reformulation.\n3. Estimate retrieval difficulty from 1–10. This controls a budget of complete\n   follow-up rounds: difficulty 1–3 gets 0 rounds, 4–7 gets 1, and 8–10 gets 2.\n4. Resolve one publication window and priority for the request. Every initial\n   and follow-up call must inherit those exact controls; never widen a window or\n   change priority during the loop. Every returned candidate already satisfies\n   that window, so rank what is available instead of lamenting well-known work\n   that the user excluded.\n\n### Run and rank\n\n1. Choose the initial sources and strategies that fit the query. For arXiv-heavy\n   ML, CS, math, or physics questions, use alphaXiv keyword, embedding, or both\n   according to whether exact full-text evidence, semantic coverage, or both are\n   useful. Add OpenAlex for broader journal, conference, citation, or\n   cross-disciplinary coverage. Use bioRxiv for biology and adjacent\n   life-science preprints, not as a ritual call for unrelated topics. When the\n   corpus is genuinely ambiguous or interdisciplinary, query multiple relevant\n   sources concurrently. If the initial round includes alphaXiv keyword and its\n   terms mix other terms with one or more 2–10 character tokens\n   that start with a letter, contain only letters, digits, or hyphens, and have\n   at least two uppercase letters, concurrently run one additional keyword call\n   whose query is exactly those acronym tokens joined by spaces and nothing\n   else. This recovery call is part of the initial round.\n2. Treat initial calls independently: retain every successful result set when\n   another call fails. If none returns results and follow-up budget remains,\n   use a round only when a focused recovery query is likely to work.\n3. Inspect and deduplicate every candidate. Match exact `id` first, then a DOI\n   or arXiv id visible in the metadata, then exact normalized title as a\n   cross-source fallback. Prefer the alphaXiv representation of an arXiv\n   duplicate because it supports full-text reading. bioRxiv is a subset of\n   OpenAlex, so overlap between those calls is expected. Within each source,\n   the API order already blends topical relevance with the requested priority:\n   - With `recency`, freshness is already upranked and old accumulated votes\n     are damped. Reorder only for topical fit; do not exclude an older but much\n     better match.\n   - With `popular`, votes or citations dominate among topically plausible\n     results. Keep high-impact relevant papers, but drop off-topic ones.\n   - Otherwise, topical relevance remains primary with freshness and votes\n     already nudging the order. Do not apply those preferences a second time.\n4. If the initial candidates provide solid topical coverage, stop immediately\n   and rank 5–15 IDs. Fast and slightly less complete is better than an\n   exploratory search. Prefer fewer strong papers over padding.\n5. Otherwise, spend at most the difficulty-derived number of follow-up rounds.\n   One round targets one concrete missing acronym, method, benchmark,\n   organization, title phrase, venue, or subtopic. Choose one or more sources\n   based on the gap: alphaXiv keyword for exact\u002Ffull-text evidence, alphaXiv\n   embedding for a semantic arXiv angle, OpenAlex for broad scholarly or\n   citation coverage, and bioRxiv for recent biology preprints. Later rounds do\n   **not** need to query all sources. Calls for the same missing angle count\n   together as one round. Never spend a round merely rephrasing an existing\n   search. Re-evaluate after each round and stop as soon as coverage is\n   sufficient. The budget is a hard cap, not a target.\n6. Drop each selected ID that did not appear in a successful initial or\n   follow-up result, retaining the surviving IDs in your chosen rank order. If\n   no selected ID survives, fall back to the first 15 unique IDs in observation\n   order, with initial results before follow-up results. Never invent or recall\n   an ID.\n\nBatch all facets into one broad retrieval loop and plan against a cap of two\ncomplete loops per user turn. If a genuinely distinct topic still forces a\nthird or fourth loop, run it in shallow mode: initial searches only, with zero\nfollow-up rounds. This degradation is a backstop, not permission to plan extra\nloops. Refuse a fifth loop and answer from the papers already found.\n\nFor a set-of-papers request such as “find papers,” “top papers,” “what is out\nthere,” or “what should I read,” return the ranked discovery results and stop.\nDepth on individual papers is not part of the discovery loop. When the request\ninstead needs claim-level synthesis, methodological details, or comparison,\nfinish retrieval first and then read the 3–5 most load-bearing candidates with\n`orx paper \u003Cid>` (or the number the user requested). Do not narrow to 3–5\npapers before retrieval has produced its ranked 5–15 candidate set.\n\nDo not compare alphaXiv votes numerically with OpenAlex citations; they measure\ndifferent things. Topical fit is the cross-source ranking signal.\n\nIn the final answer, link every alphaXiv\u002FarXiv paper title or paper ID to\n`https:\u002F\u002Fwww.alphaxiv.org\u002Fabs\u002F\u003Cversionless-paperId>`. Never return an\n`arxiv.org` link for those papers. Link a DOI result to `https:\u002F\u002Fdoi.org\u002F\u003Cdoi>`\nand a bare OpenAlex `W…` id to `https:\u002F\u002Fopenalex.org\u002F\u003Cid>`.\n\nFor claim-level synthesis, place the supporting source link immediately after\neach substantive scholarly claim, and use a paper as claim-level support only\nafter reading it. A discovery-only result list may link candidate titles, but\nmust not imply that their methods or findings were verified from snippets alone.\n\n## Reading selected papers\n\n`orx paper` auto-detects an arXiv id\u002FURL, bioRxiv DOI, other DOI, or OpenAlex\n`W…` id. For alphaXiv it returns a compact structured report; use `--full` only\nwhen you explicitly need raw text even if a report exists. Without `--full`, a\nmissing report automatically falls back to extracted full text in the same\ncommand. `--full` skips the report entirely rather than acting as a superset of\nthe default. If extracted text is also unavailable, use the alphaXiv paper link\nit returns.\n\n`orx paper` prints the alphaXiv link before the content. When alphaXiv has an\nassociated repository, it then prints `GitHub: \u003Curl>`. This is the most-starred\nassociated repository and can be a framework rather than the paper's own code,\nso sanity-check it before treating it as the implementation.\n\nAll discovery and paper commands honor the user's disabled literature-source\nsettings; do not work around an error saying a source is disabled.\n",{"name":5,"description":7},"main","d69096e239f2a89aabc2cd078c47664a9d3b1e11","07-后端与架构\u002FalphaXiv__OpenResearch\u002Fagent-skills\u002Forx-lit-review",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":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add lza6\u002FArmor-piercing-system-prompts --skill orx-lit-review",{"tool":38,"command":39},"bunx","bunx skills add lza6\u002FArmor-piercing-system-prompts --skill orx-lit-review",{"tool":41,"command":42},"pnpm","pnpm dlx skills add lza6\u002FArmor-piercing-system-prompts --skill orx-lit-review","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002FalphaXiv__OpenResearch\u002Fagent-skills\u002Forx-lit-review\u002FSKILL.md",true,[],[47,54,59,66,73,80],{"skillNo":48,"name":49,"nameZh":6,"description":50,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":51,"categories":52,"keywords":53,"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.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobsidian-clipper-template-creator\u002FSKILL.md",[],[],{"skillNo":55,"name":49,"nameZh":6,"description":50,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":56,"categories":57,"keywords":58,"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":60,"name":61,"nameZh":6,"description":62,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":63,"categories":64,"keywords":65,"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":67,"name":68,"nameZh":6,"description":69,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":72,"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":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"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":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"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",[],[]]