[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-huggingface-pwc-cli":3,"mdc--fcdl8f-key":32,"related-repo-huggingface-pwc-cli":964,"related-org-huggingface-pwc-cli":972},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":21,"repoUrl":22,"updatedAt":23,"license":24,"forks":25,"topics":26,"repo":27,"sourceUrl":30,"mdContent":31},"pwc-cli","research papers and benchmarks","Papers With Code CLI (`pwc`) for searching and reading AI\u002FML papers, discovering recent and trending research, finding related work and paper lineage, browsing tasks, methods, conferences, and benchmark leaderboards through the public Papers With Code catalog. Use whenever the user asks to find papers, survey literature, compare research, inspect an arXiv paper, explore AI\u002FML taxonomy or conferences, discover benchmarks or state-of-the-art models, or mentions Papers With Code, `pwc`, or the `pwc-cli`. Prefer this skill for grounded AI\u002FML research even when the user does not explicitly ask for a CLI command.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},"huggingface","Hugging Face","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fhuggingface.png",[12,16,18],{"name":13,"slug":14,"type":15},"Research","research","tag",{"name":9,"slug":17,"type":15},"hugging-face",{"name":19,"slug":20,"type":15},"CLI","cli",1,"https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Fpwc-cli","2026-07-31T05:52:04.168908",null,0,[],{"repoUrl":22,"stars":21,"forks":25,"topics":28,"description":29},[],"A Command-Line Interface (CLI) to interact with Papers with Code. For humans and agents.","https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Fpwc-cli\u002Ftree\u002FHEAD\u002Fstandalone_cli","---\nname: pwc-cli\ndescription: \"Papers With Code CLI (`pwc`) for searching and reading AI\u002FML papers, discovering recent and trending research, finding related work and paper lineage, browsing tasks, methods, conferences, and benchmark leaderboards through the public Papers With Code catalog. Use whenever the user asks to find papers, survey literature, compare research, inspect an arXiv paper, explore AI\u002FML taxonomy or conferences, discover benchmarks or state-of-the-art models, or mentions Papers With Code, `pwc`, or the `pwc-cli`. Prefer this skill for grounded AI\u002FML research even when the user does not explicitly ask for a CLI command.\"\n---\n\nInstall:\n\n```bash\npwc_repo=https:\u002F\u002Fraw.githubusercontent.com\u002Fhuggingface\u002Fpwc-cli\u002Fmain\ncurl -LsSf \"$pwc_repo\u002Fstandalone_cli\u002Finstall.py\" | python3\n```\n\nThe `pwc` CLI is anonymous and read-only. It queries the public\n[Papers With Code](https:\u002F\u002Fpaperswithcode.co) catalog and requires no token.\nRun `pwc --help` or a nested `--help` command when the live parser and this skill\ndisagree; the parser is authoritative.\n\nUse compact output for reading and discovery. Add `--json` when programmatic\nfiltering, joining, or schema-dependent processing is useful.\nUse only flags documented for the exact subcommand; do not infer that sibling\ncommands share flags. For `pwc paper list`, `pwc task list`, and\n`pwc method list`, use `--page-size`, never `--limit`.\n\n`PAPER` accepts a modern or legacy arXiv ID, a numeric external-paper ID, or an\nexact paper title. Papers with Code supports external papers which are not on arXiv,\nhence those papers will have a numeric canonical ID.\nQuote titles containing spaces. Title matching is\ncase-insensitive but exact; ambiguous titles fail with their matching IDs rather\nthan silently selecting a paper.\n\n## Commands\n\n- `pwc search QUERY` — Search papers by title, topic, author, or arXiv ID. This is powered by hybrid\u002Fkeyword\u002Fsemantic search using pgvector. Defaults to hybrid.\n  `[--limit 1-100 --page 1-100 --mode hybrid|keyword|semantic --json]`\n- `pwc paper info PAPER` — Show metadata, including the abstract.\n  `--include-resources` adds Markdown sections for GitHub repositories, project pages,\n  and Hugging Face artifacts, with official links marked explicitly.\n  `[--include-resources --json]`\n- `pwc paper read PAPER` — Print stored paper Markdown. The resolved paper must\n  have a modern arXiv record. `[--json]`\n- `pwc paper list` — List and filter the paper catalog in a paginated manner.\n  `[--page 1-100 --page-size 1-100 --search TEXT\n  --published-after YYYY-MM-DD --published-before YYYY-MM-DD\n  --conference TEXT --all-versions\n  --order-by trending|date_published|citation_count\n  --order-dir asc|desc --time today|week|month|all_time\n  --include-resources --json]`\n- `pwc paper recent` — List recently published papers.\n  `[--limit 1-100 --json]`\n- `pwc paper trending` — List papers with recent repository activity.\n  `[--limit 1-100 --max-age-days 1-365 --min-velocity FLOAT --json]`\n- `pwc paper related PAPER` — Find embedding- and taxonomy-ranked related work.\n  `[--limit 1-20 --json]`\n- `pwc paper lineage list PAPER` — Render linked Markdown sections for a paper,\n  its predecessors, and its successors. `[--json]`\n- `pwc task list` — List research tasks, optionally filtering by a\n  case-insensitive exact area name such as Vision, Audio, or General.\n  Interactive output groups visible top-level tasks into Markdown area sections;\n  `--group-by-area` forces that view when output is captured, while `--flat`\n  forces the paginated table.\n  `[--page 1-100 --page-size 1-100 --area NAME_OR_ID --level INTEGER\n  --visible-only --group-by-area|--flat\n  --order-by name|created_at|level|paper_count\n  --order-dir asc|desc --json]`\n- `pwc method list` — List research methods, optionally filtering by area or\n  introduction year. This subcommand has no `--search` or `--limit` flag; do\n  not infer flags from sibling list commands. To find methods by name or topic,\n  filter the bounded list with\n  `pwc method list --page-size 500 | rg -i -- 'QUERY'`.\n  `[--page 1-100 --page-size 1-500 --area NAME_OR_ID\n  --introduced-year YEAR\n  --order-by name|full_name|introduced_year|created_at|paper_count\n  --order-dir asc|desc --json]`\n- `pwc conference list` — List conferences with imported papers.\n  `[--year YEAR --json]`\n- `pwc benchmark list` — Find and rank benchmarks. When `--task` is supplied,\n  the default order is task-specific trending activity. Without filters,\n  interactive output groups top benchmarks under visible tasks by area;\n  `--group-by-area` forces Markdown and `--flat` forces the dataset table.\n  Task, search, pagination, ordering, and open-model filters take precedence\n  over `--group-by-area` and select task-scoped flat output.\n  Rows include the benchmark name, slug, and ID (plus the full name when\n  distinct in grouped output); pass the slug or ID to\n  `pwc benchmark --name IDENTIFIER` for leaderboard details.\n  `[--page 1-100 --page-size 1-100 --search TEXT --task TASK --area NAME_OR_ID\n  --benchmarks-per-task 1-10 --group-by-area|--flat\n  --include-descendants --min-eval-count INTEGER --is-open true|false\n  --order-by trending|name|full_name|created_at|paper_count\n  --order-dir asc|desc --json]`\n- `pwc benchmark --name NAME` — Show a benchmark's top models, scores, source\n  papers, publication dates, and open\u002Fclosed status. `NAME` must exactly match\n  a benchmark name, full name, slug, or ID.\n  For multi-metric tradeoffs, require numeric metrics, apply repeatable\n  thresholds, sort by one metric, or select a Pareto frontier. Metric names are\n  case-insensitive.\n  `[--limit 1-100 --is-open true|false\n  --require-metrics METRIC[,METRIC]\n  --min METRIC=VALUE --max METRIC=VALUE\n  --sort METRIC[:asc|desc]\n  --pareto METRIC:higher,METRIC:lower --json]`\n- `pwc version` — Show the CLI and API contract versions.\n\n## Research workflow\n\n1. Use `pwc benchmark list --task TASK` to discover active benchmarks for a given task,\n   then `pwc benchmark --name NAME` to inspect a specific leaderboard.\n   For accuracy\u002Flatency or other multi-metric questions, use\n   `--require-metrics`, thresholds, or `--pareto`; do not infer a tradeoff from\n   a leaderboard sorted by one metric.\n2. Use `pwc paper info` to inspect promising results. Add `--include-resources`\n   when linked GitHub repositories, project pages, or Hugging Face artifacts matter.\n3. Use `pwc search` to search more broadly for a topic or a known paper.\n   Use the default hybrid mode first; use keyword mode for exact terminology\n   and semantic mode for conceptual matches.\n4. Go more in-depth with `pwc paper read`. Do not treat search snippets or\n   titles as sufficient support for detailed claims.\n5. Expand the literature with `pwc paper related` and use\n   `pwc paper lineage list` when model or method ancestry matters.\n6. Explore the catalog taxonomy with `pwc task list --group-by-area`,\n   `pwc method list`, and `pwc conference list`; use `--area` or `--year` to\n   narrow broad lists.\n7. Synthesize only after gathering enough primary evidence. Preserve paper\n   titles, identifiers, and URLs in the answer so claims remain traceable.\n\n## Command selection\n\n- Use `pwc benchmark list --task TASK` for finding state-of-the-art (SOTA) for a given task.\n- Use `pwc search` for relevance-ranked discovery.\n- Use `pwc paper list` for structured filters, date windows, conferences, and\n  deterministic sorting.\n- Use `pwc paper recent` for recency and `pwc paper trending` for current\n  repository activity; these are different signals.\n- Use area names directly with task and method lists, for example\n  `--area Vision`, `--area Audio`, or `--area General`. Area matching is\n  case-insensitive; numeric area IDs are also accepted.\n- Use `pwc conference list --year YEAR` for a specific conference edition year.\n- Use `--include-resources` for linked repositories, project pages, and Hugging\n  Face artifacts. Add `--json` when their structured metadata also matters.\n- Use `--all-versions` only when individual arXiv versions are relevant.\n- Use `--is-open true` to restrict benchmark results to open models.\n- Paginate when stderr reports more results. Do not silently treat the first\n  page as the complete catalog.\n\n## Output and limits\n\n- Interactive list and search output uses aligned text columns. When stdout is\n  piped or captured, the same rows use lossless TSV for agent and script\n  processing. Paper info uses labeled metadata and Markdown sections; benchmark\n  details use Markdown.\n- `--json` wraps responses as\n  `{\"schema_version\":\"v1\",\"data\":...}` for stable agent consumption.\n- Stable exit codes are `0` success, `2` invalid usage, `3` network\u002Fserver\n  failure, and `4` invalid API response.\n- `PWC_API_URL` may select another compatible v1 endpoint. The default is\n  `https:\u002F\u002Fpaperswithcode.co\u002Fapi\u002Fv1`.\n\nThe standalone parser contains no authentication, mutation, ingestion,\npublication, image, CRON, embedding, or infrastructure commands. Do not\nsubstitute repository-maintenance commands into a public or sandbox workflow.\n",{"data":33,"body":34},{"name":4,"description":6},{"type":35,"children":36},"root",[37,45,125,165,218,229,236,561,567,719,725,866,872,953,958],{"type":38,"tag":39,"props":40,"children":41},"element","p",{},[42],{"type":43,"value":44},"text","Install:",{"type":38,"tag":46,"props":47,"children":52},"pre",{"className":48,"code":49,"language":50,"meta":51,"style":51},"language-bash shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","pwc_repo=https:\u002F\u002Fraw.githubusercontent.com\u002Fhuggingface\u002Fpwc-cli\u002Fmain\ncurl -LsSf \"$pwc_repo\u002Fstandalone_cli\u002Finstall.py\" | python3\n","bash","",[53],{"type":38,"tag":54,"props":55,"children":56},"code",{"__ignoreMap":51},[57,80],{"type":38,"tag":58,"props":59,"children":61},"span",{"class":60,"line":21},"line",[62,68,74],{"type":38,"tag":58,"props":63,"children":65},{"style":64},"--shiki-light:#90A4AE;--shiki-default:#EEFFFF;--shiki-dark:#BABED8",[66],{"type":43,"value":67},"pwc_repo",{"type":38,"tag":58,"props":69,"children":71},{"style":70},"--shiki-light:#39ADB5;--shiki-default:#89DDFF;--shiki-dark:#89DDFF",[72],{"type":43,"value":73},"=",{"type":38,"tag":58,"props":75,"children":77},{"style":76},"--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D",[78],{"type":43,"value":79},"https:\u002F\u002Fraw.githubusercontent.com\u002Fhuggingface\u002Fpwc-cli\u002Fmain\n",{"type":38,"tag":58,"props":81,"children":83},{"class":60,"line":82},2,[84,90,95,100,105,110,115,120],{"type":38,"tag":58,"props":85,"children":87},{"style":86},"--shiki-light:#E2931D;--shiki-default:#FFCB6B;--shiki-dark:#FFCB6B",[88],{"type":43,"value":89},"curl",{"type":38,"tag":58,"props":91,"children":92},{"style":76},[93],{"type":43,"value":94}," -LsSf",{"type":38,"tag":58,"props":96,"children":97},{"style":70},[98],{"type":43,"value":99}," \"",{"type":38,"tag":58,"props":101,"children":102},{"style":64},[103],{"type":43,"value":104},"$pwc_repo",{"type":38,"tag":58,"props":106,"children":107},{"style":76},[108],{"type":43,"value":109},"\u002Fstandalone_cli\u002Finstall.py",{"type":38,"tag":58,"props":111,"children":112},{"style":70},[113],{"type":43,"value":114},"\"",{"type":38,"tag":58,"props":116,"children":117},{"style":70},[118],{"type":43,"value":119}," |",{"type":38,"tag":58,"props":121,"children":122},{"style":86},[123],{"type":43,"value":124}," python3\n",{"type":38,"tag":39,"props":126,"children":127},{},[128,130,136,138,147,149,155,157,163],{"type":43,"value":129},"The ",{"type":38,"tag":54,"props":131,"children":133},{"className":132},[],[134],{"type":43,"value":135},"pwc",{"type":43,"value":137}," CLI is anonymous and read-only. It queries the public\n",{"type":38,"tag":139,"props":140,"children":144},"a",{"href":141,"rel":142},"https:\u002F\u002Fpaperswithcode.co",[143],"nofollow",[145],{"type":43,"value":146},"Papers With Code",{"type":43,"value":148}," catalog and requires no token.\nRun ",{"type":38,"tag":54,"props":150,"children":152},{"className":151},[],[153],{"type":43,"value":154},"pwc --help",{"type":43,"value":156}," or a nested ",{"type":38,"tag":54,"props":158,"children":160},{"className":159},[],[161],{"type":43,"value":162},"--help",{"type":43,"value":164}," command when the live parser and this skill\ndisagree; the parser is authoritative.",{"type":38,"tag":39,"props":166,"children":167},{},[168,170,176,178,184,186,192,194,200,202,208,210,216],{"type":43,"value":169},"Use compact output for reading and discovery. Add ",{"type":38,"tag":54,"props":171,"children":173},{"className":172},[],[174],{"type":43,"value":175},"--json",{"type":43,"value":177}," when programmatic\nfiltering, joining, or schema-dependent processing is useful.\nUse only flags documented for the exact subcommand; do not infer that sibling\ncommands share flags. For ",{"type":38,"tag":54,"props":179,"children":181},{"className":180},[],[182],{"type":43,"value":183},"pwc paper list",{"type":43,"value":185},", ",{"type":38,"tag":54,"props":187,"children":189},{"className":188},[],[190],{"type":43,"value":191},"pwc task list",{"type":43,"value":193},", and\n",{"type":38,"tag":54,"props":195,"children":197},{"className":196},[],[198],{"type":43,"value":199},"pwc method list",{"type":43,"value":201},", use ",{"type":38,"tag":54,"props":203,"children":205},{"className":204},[],[206],{"type":43,"value":207},"--page-size",{"type":43,"value":209},", never ",{"type":38,"tag":54,"props":211,"children":213},{"className":212},[],[214],{"type":43,"value":215},"--limit",{"type":43,"value":217},".",{"type":38,"tag":39,"props":219,"children":220},{},[221,227],{"type":38,"tag":54,"props":222,"children":224},{"className":223},[],[225],{"type":43,"value":226},"PAPER",{"type":43,"value":228}," accepts a modern or legacy arXiv ID, a numeric external-paper ID, or an\nexact paper title. Papers with Code supports external papers which are not on arXiv,\nhence those papers will have a numeric canonical ID.\nQuote titles containing spaces. Title matching is\ncase-insensitive but exact; ambiguous titles fail with their matching IDs rather\nthan silently selecting a paper.",{"type":38,"tag":230,"props":231,"children":233},"h2",{"id":232},"commands",[234],{"type":43,"value":235},"Commands",{"type":38,"tag":237,"props":238,"children":239},"ul",{},[240,258,283,300,316,333,350,367,383,415,454,471,525,550],{"type":38,"tag":241,"props":242,"children":243},"li",{},[244,250,252],{"type":38,"tag":54,"props":245,"children":247},{"className":246},[],[248],{"type":43,"value":249},"pwc search QUERY",{"type":43,"value":251}," — Search papers by title, topic, author, or arXiv ID. This is powered by hybrid\u002Fkeyword\u002Fsemantic search using pgvector. Defaults to hybrid.\n",{"type":38,"tag":54,"props":253,"children":255},{"className":254},[],[256],{"type":43,"value":257},"[--limit 1-100 --page 1-100 --mode hybrid|keyword|semantic --json]",{"type":38,"tag":241,"props":259,"children":260},{},[261,267,269,275,277],{"type":38,"tag":54,"props":262,"children":264},{"className":263},[],[265],{"type":43,"value":266},"pwc paper info PAPER",{"type":43,"value":268}," — Show metadata, including the abstract.\n",{"type":38,"tag":54,"props":270,"children":272},{"className":271},[],[273],{"type":43,"value":274},"--include-resources",{"type":43,"value":276}," adds Markdown sections for GitHub repositories, project pages,\nand Hugging Face artifacts, with official links marked explicitly.\n",{"type":38,"tag":54,"props":278,"children":280},{"className":279},[],[281],{"type":43,"value":282},"[--include-resources --json]",{"type":38,"tag":241,"props":284,"children":285},{},[286,292,294],{"type":38,"tag":54,"props":287,"children":289},{"className":288},[],[290],{"type":43,"value":291},"pwc paper read PAPER",{"type":43,"value":293}," — Print stored paper Markdown. The resolved paper must\nhave a modern arXiv record. ",{"type":38,"tag":54,"props":295,"children":297},{"className":296},[],[298],{"type":43,"value":299},"[--json]",{"type":38,"tag":241,"props":301,"children":302},{},[303,308,310],{"type":38,"tag":54,"props":304,"children":306},{"className":305},[],[307],{"type":43,"value":183},{"type":43,"value":309}," — List and filter the paper catalog in a paginated manner.\n",{"type":38,"tag":54,"props":311,"children":313},{"className":312},[],[314],{"type":43,"value":315},"[--page 1-100 --page-size 1-100 --search TEXT --published-after YYYY-MM-DD --published-before YYYY-MM-DD --conference TEXT --all-versions --order-by trending|date_published|citation_count --order-dir asc|desc --time today|week|month|all_time --include-resources --json]",{"type":38,"tag":241,"props":317,"children":318},{},[319,325,327],{"type":38,"tag":54,"props":320,"children":322},{"className":321},[],[323],{"type":43,"value":324},"pwc paper recent",{"type":43,"value":326}," — List recently published papers.\n",{"type":38,"tag":54,"props":328,"children":330},{"className":329},[],[331],{"type":43,"value":332},"[--limit 1-100 --json]",{"type":38,"tag":241,"props":334,"children":335},{},[336,342,344],{"type":38,"tag":54,"props":337,"children":339},{"className":338},[],[340],{"type":43,"value":341},"pwc paper trending",{"type":43,"value":343}," — List papers with recent repository activity.\n",{"type":38,"tag":54,"props":345,"children":347},{"className":346},[],[348],{"type":43,"value":349},"[--limit 1-100 --max-age-days 1-365 --min-velocity FLOAT --json]",{"type":38,"tag":241,"props":351,"children":352},{},[353,359,361],{"type":38,"tag":54,"props":354,"children":356},{"className":355},[],[357],{"type":43,"value":358},"pwc paper related PAPER",{"type":43,"value":360}," — Find embedding- and taxonomy-ranked related work.\n",{"type":38,"tag":54,"props":362,"children":364},{"className":363},[],[365],{"type":43,"value":366},"[--limit 1-20 --json]",{"type":38,"tag":241,"props":368,"children":369},{},[370,376,378],{"type":38,"tag":54,"props":371,"children":373},{"className":372},[],[374],{"type":43,"value":375},"pwc paper lineage list PAPER",{"type":43,"value":377}," — Render linked Markdown sections for a paper,\nits predecessors, and its successors. ",{"type":38,"tag":54,"props":379,"children":381},{"className":380},[],[382],{"type":43,"value":299},{"type":38,"tag":241,"props":384,"children":385},{},[386,391,393,399,401,407,409],{"type":38,"tag":54,"props":387,"children":389},{"className":388},[],[390],{"type":43,"value":191},{"type":43,"value":392}," — List research tasks, optionally filtering by a\ncase-insensitive exact area name such as Vision, Audio, or General.\nInteractive output groups visible top-level tasks into Markdown area sections;\n",{"type":38,"tag":54,"props":394,"children":396},{"className":395},[],[397],{"type":43,"value":398},"--group-by-area",{"type":43,"value":400}," forces that view when output is captured, while ",{"type":38,"tag":54,"props":402,"children":404},{"className":403},[],[405],{"type":43,"value":406},"--flat",{"type":43,"value":408},"\nforces the paginated table.\n",{"type":38,"tag":54,"props":410,"children":412},{"className":411},[],[413],{"type":43,"value":414},"[--page 1-100 --page-size 1-100 --area NAME_OR_ID --level INTEGER --visible-only --group-by-area|--flat --order-by name|created_at|level|paper_count --order-dir asc|desc --json]",{"type":38,"tag":241,"props":416,"children":417},{},[418,423,425,431,433,438,440,446,448],{"type":38,"tag":54,"props":419,"children":421},{"className":420},[],[422],{"type":43,"value":199},{"type":43,"value":424}," — List research methods, optionally filtering by area or\nintroduction year. This subcommand has no ",{"type":38,"tag":54,"props":426,"children":428},{"className":427},[],[429],{"type":43,"value":430},"--search",{"type":43,"value":432}," or ",{"type":38,"tag":54,"props":434,"children":436},{"className":435},[],[437],{"type":43,"value":215},{"type":43,"value":439}," flag; do\nnot infer flags from sibling list commands. To find methods by name or topic,\nfilter the bounded list with\n",{"type":38,"tag":54,"props":441,"children":443},{"className":442},[],[444],{"type":43,"value":445},"pwc method list --page-size 500 | rg -i -- 'QUERY'",{"type":43,"value":447},".\n",{"type":38,"tag":54,"props":449,"children":451},{"className":450},[],[452],{"type":43,"value":453},"[--page 1-100 --page-size 1-500 --area NAME_OR_ID --introduced-year YEAR --order-by name|full_name|introduced_year|created_at|paper_count --order-dir asc|desc --json]",{"type":38,"tag":241,"props":455,"children":456},{},[457,463,465],{"type":38,"tag":54,"props":458,"children":460},{"className":459},[],[461],{"type":43,"value":462},"pwc conference list",{"type":43,"value":464}," — List conferences with imported papers.\n",{"type":38,"tag":54,"props":466,"children":468},{"className":467},[],[469],{"type":43,"value":470},"[--year YEAR --json]",{"type":38,"tag":241,"props":472,"children":473},{},[474,480,482,488,490,495,497,502,504,509,511,517,519],{"type":38,"tag":54,"props":475,"children":477},{"className":476},[],[478],{"type":43,"value":479},"pwc benchmark list",{"type":43,"value":481}," — Find and rank benchmarks. When ",{"type":38,"tag":54,"props":483,"children":485},{"className":484},[],[486],{"type":43,"value":487},"--task",{"type":43,"value":489}," is supplied,\nthe default order is task-specific trending activity. Without filters,\ninteractive output groups top benchmarks under visible tasks by area;\n",{"type":38,"tag":54,"props":491,"children":493},{"className":492},[],[494],{"type":43,"value":398},{"type":43,"value":496}," forces Markdown and ",{"type":38,"tag":54,"props":498,"children":500},{"className":499},[],[501],{"type":43,"value":406},{"type":43,"value":503}," forces the dataset table.\nTask, search, pagination, ordering, and open-model filters take precedence\nover ",{"type":38,"tag":54,"props":505,"children":507},{"className":506},[],[508],{"type":43,"value":398},{"type":43,"value":510}," and select task-scoped flat output.\nRows include the benchmark name, slug, and ID (plus the full name when\ndistinct in grouped output); pass the slug or ID to\n",{"type":38,"tag":54,"props":512,"children":514},{"className":513},[],[515],{"type":43,"value":516},"pwc benchmark --name IDENTIFIER",{"type":43,"value":518}," for leaderboard details.\n",{"type":38,"tag":54,"props":520,"children":522},{"className":521},[],[523],{"type":43,"value":524},"[--page 1-100 --page-size 1-100 --search TEXT --task TASK --area NAME_OR_ID --benchmarks-per-task 1-10 --group-by-area|--flat --include-descendants --min-eval-count INTEGER --is-open true|false --order-by trending|name|full_name|created_at|paper_count --order-dir asc|desc --json]",{"type":38,"tag":241,"props":526,"children":527},{},[528,534,536,542,544],{"type":38,"tag":54,"props":529,"children":531},{"className":530},[],[532],{"type":43,"value":533},"pwc benchmark --name NAME",{"type":43,"value":535}," — Show a benchmark's top models, scores, source\npapers, publication dates, and open\u002Fclosed status. ",{"type":38,"tag":54,"props":537,"children":539},{"className":538},[],[540],{"type":43,"value":541},"NAME",{"type":43,"value":543}," must exactly match\na benchmark name, full name, slug, or ID.\nFor multi-metric tradeoffs, require numeric metrics, apply repeatable\nthresholds, sort by one metric, or select a Pareto frontier. Metric names are\ncase-insensitive.\n",{"type":38,"tag":54,"props":545,"children":547},{"className":546},[],[548],{"type":43,"value":549},"[--limit 1-100 --is-open true|false --require-metrics METRIC[,METRIC] --min METRIC=VALUE --max METRIC=VALUE --sort METRIC[:asc|desc] --pareto METRIC:higher,METRIC:lower --json]",{"type":38,"tag":241,"props":551,"children":552},{},[553,559],{"type":38,"tag":54,"props":554,"children":556},{"className":555},[],[557],{"type":43,"value":558},"pwc version",{"type":43,"value":560}," — Show the CLI and API contract versions.",{"type":38,"tag":230,"props":562,"children":564},{"id":563},"research-workflow",[565],{"type":43,"value":566},"Research workflow",{"type":38,"tag":568,"props":569,"children":570},"ol",{},[571,607,626,638,651,672,714],{"type":38,"tag":241,"props":572,"children":573},{},[574,576,582,584,589,591,597,599,605],{"type":43,"value":575},"Use ",{"type":38,"tag":54,"props":577,"children":579},{"className":578},[],[580],{"type":43,"value":581},"pwc benchmark list --task TASK",{"type":43,"value":583}," to discover active benchmarks for a given task,\nthen ",{"type":38,"tag":54,"props":585,"children":587},{"className":586},[],[588],{"type":43,"value":533},{"type":43,"value":590}," to inspect a specific leaderboard.\nFor accuracy\u002Flatency or other multi-metric questions, use\n",{"type":38,"tag":54,"props":592,"children":594},{"className":593},[],[595],{"type":43,"value":596},"--require-metrics",{"type":43,"value":598},", thresholds, or ",{"type":38,"tag":54,"props":600,"children":602},{"className":601},[],[603],{"type":43,"value":604},"--pareto",{"type":43,"value":606},"; do not infer a tradeoff from\na leaderboard sorted by one metric.",{"type":38,"tag":241,"props":608,"children":609},{},[610,611,617,619,624],{"type":43,"value":575},{"type":38,"tag":54,"props":612,"children":614},{"className":613},[],[615],{"type":43,"value":616},"pwc paper info",{"type":43,"value":618}," to inspect promising results. Add ",{"type":38,"tag":54,"props":620,"children":622},{"className":621},[],[623],{"type":43,"value":274},{"type":43,"value":625},"\nwhen linked GitHub repositories, project pages, or Hugging Face artifacts matter.",{"type":38,"tag":241,"props":627,"children":628},{},[629,630,636],{"type":43,"value":575},{"type":38,"tag":54,"props":631,"children":633},{"className":632},[],[634],{"type":43,"value":635},"pwc search",{"type":43,"value":637}," to search more broadly for a topic or a known paper.\nUse the default hybrid mode first; use keyword mode for exact terminology\nand semantic mode for conceptual matches.",{"type":38,"tag":241,"props":639,"children":640},{},[641,643,649],{"type":43,"value":642},"Go more in-depth with ",{"type":38,"tag":54,"props":644,"children":646},{"className":645},[],[647],{"type":43,"value":648},"pwc paper read",{"type":43,"value":650},". Do not treat search snippets or\ntitles as sufficient support for detailed claims.",{"type":38,"tag":241,"props":652,"children":653},{},[654,656,662,664,670],{"type":43,"value":655},"Expand the literature with ",{"type":38,"tag":54,"props":657,"children":659},{"className":658},[],[660],{"type":43,"value":661},"pwc paper related",{"type":43,"value":663}," and use\n",{"type":38,"tag":54,"props":665,"children":667},{"className":666},[],[668],{"type":43,"value":669},"pwc paper lineage list",{"type":43,"value":671}," when model or method ancestry matters.",{"type":38,"tag":241,"props":673,"children":674},{},[675,677,683,685,690,692,697,699,705,706,712],{"type":43,"value":676},"Explore the catalog taxonomy with ",{"type":38,"tag":54,"props":678,"children":680},{"className":679},[],[681],{"type":43,"value":682},"pwc task list --group-by-area",{"type":43,"value":684},",\n",{"type":38,"tag":54,"props":686,"children":688},{"className":687},[],[689],{"type":43,"value":199},{"type":43,"value":691},", and ",{"type":38,"tag":54,"props":693,"children":695},{"className":694},[],[696],{"type":43,"value":462},{"type":43,"value":698},"; use ",{"type":38,"tag":54,"props":700,"children":702},{"className":701},[],[703],{"type":43,"value":704},"--area",{"type":43,"value":432},{"type":38,"tag":54,"props":707,"children":709},{"className":708},[],[710],{"type":43,"value":711},"--year",{"type":43,"value":713}," to\nnarrow broad lists.",{"type":38,"tag":241,"props":715,"children":716},{},[717],{"type":43,"value":718},"Synthesize only after gathering enough primary evidence. Preserve paper\ntitles, identifiers, and URLs in the answer so claims remain traceable.",{"type":38,"tag":230,"props":720,"children":722},{"id":721},"command-selection",[723],{"type":43,"value":724},"Command selection",{"type":38,"tag":237,"props":726,"children":727},{},[728,739,750,761,779,807,819,837,849,861],{"type":38,"tag":241,"props":729,"children":730},{},[731,732,737],{"type":43,"value":575},{"type":38,"tag":54,"props":733,"children":735},{"className":734},[],[736],{"type":43,"value":581},{"type":43,"value":738}," for finding state-of-the-art (SOTA) for a given task.",{"type":38,"tag":241,"props":740,"children":741},{},[742,743,748],{"type":43,"value":575},{"type":38,"tag":54,"props":744,"children":746},{"className":745},[],[747],{"type":43,"value":635},{"type":43,"value":749}," for relevance-ranked discovery.",{"type":38,"tag":241,"props":751,"children":752},{},[753,754,759],{"type":43,"value":575},{"type":38,"tag":54,"props":755,"children":757},{"className":756},[],[758],{"type":43,"value":183},{"type":43,"value":760}," for structured filters, date windows, conferences, and\ndeterministic sorting.",{"type":38,"tag":241,"props":762,"children":763},{},[764,765,770,772,777],{"type":43,"value":575},{"type":38,"tag":54,"props":766,"children":768},{"className":767},[],[769],{"type":43,"value":324},{"type":43,"value":771}," for recency and ",{"type":38,"tag":54,"props":773,"children":775},{"className":774},[],[776],{"type":43,"value":341},{"type":43,"value":778}," for current\nrepository activity; these are different signals.",{"type":38,"tag":241,"props":780,"children":781},{},[782,784,790,791,797,799,805],{"type":43,"value":783},"Use area names directly with task and method lists, for example\n",{"type":38,"tag":54,"props":785,"children":787},{"className":786},[],[788],{"type":43,"value":789},"--area Vision",{"type":43,"value":185},{"type":38,"tag":54,"props":792,"children":794},{"className":793},[],[795],{"type":43,"value":796},"--area Audio",{"type":43,"value":798},", or ",{"type":38,"tag":54,"props":800,"children":802},{"className":801},[],[803],{"type":43,"value":804},"--area General",{"type":43,"value":806},". Area matching is\ncase-insensitive; numeric area IDs are also accepted.",{"type":38,"tag":241,"props":808,"children":809},{},[810,811,817],{"type":43,"value":575},{"type":38,"tag":54,"props":812,"children":814},{"className":813},[],[815],{"type":43,"value":816},"pwc conference list --year YEAR",{"type":43,"value":818}," for a specific conference edition year.",{"type":38,"tag":241,"props":820,"children":821},{},[822,823,828,830,835],{"type":43,"value":575},{"type":38,"tag":54,"props":824,"children":826},{"className":825},[],[827],{"type":43,"value":274},{"type":43,"value":829}," for linked repositories, project pages, and Hugging\nFace artifacts. Add ",{"type":38,"tag":54,"props":831,"children":833},{"className":832},[],[834],{"type":43,"value":175},{"type":43,"value":836}," when their structured metadata also matters.",{"type":38,"tag":241,"props":838,"children":839},{},[840,841,847],{"type":43,"value":575},{"type":38,"tag":54,"props":842,"children":844},{"className":843},[],[845],{"type":43,"value":846},"--all-versions",{"type":43,"value":848}," only when individual arXiv versions are relevant.",{"type":38,"tag":241,"props":850,"children":851},{},[852,853,859],{"type":43,"value":575},{"type":38,"tag":54,"props":854,"children":856},{"className":855},[],[857],{"type":43,"value":858},"--is-open true",{"type":43,"value":860}," to restrict benchmark results to open models.",{"type":38,"tag":241,"props":862,"children":863},{},[864],{"type":43,"value":865},"Paginate when stderr reports more results. Do not silently treat the first\npage as the complete catalog.",{"type":38,"tag":230,"props":867,"children":869},{"id":868},"output-and-limits",[870],{"type":43,"value":871},"Output and limits",{"type":38,"tag":237,"props":873,"children":874},{},[875,880,898,935],{"type":38,"tag":241,"props":876,"children":877},{},[878],{"type":43,"value":879},"Interactive list and search output uses aligned text columns. When stdout is\npiped or captured, the same rows use lossless TSV for agent and script\nprocessing. Paper info uses labeled metadata and Markdown sections; benchmark\ndetails use Markdown.",{"type":38,"tag":241,"props":881,"children":882},{},[883,888,890,896],{"type":38,"tag":54,"props":884,"children":886},{"className":885},[],[887],{"type":43,"value":175},{"type":43,"value":889}," wraps responses as\n",{"type":38,"tag":54,"props":891,"children":893},{"className":892},[],[894],{"type":43,"value":895},"{\"schema_version\":\"v1\",\"data\":...}",{"type":43,"value":897}," for stable agent consumption.",{"type":38,"tag":241,"props":899,"children":900},{},[901,903,909,911,917,919,925,927,933],{"type":43,"value":902},"Stable exit codes are ",{"type":38,"tag":54,"props":904,"children":906},{"className":905},[],[907],{"type":43,"value":908},"0",{"type":43,"value":910}," success, ",{"type":38,"tag":54,"props":912,"children":914},{"className":913},[],[915],{"type":43,"value":916},"2",{"type":43,"value":918}," invalid usage, ",{"type":38,"tag":54,"props":920,"children":922},{"className":921},[],[923],{"type":43,"value":924},"3",{"type":43,"value":926}," network\u002Fserver\nfailure, and ",{"type":38,"tag":54,"props":928,"children":930},{"className":929},[],[931],{"type":43,"value":932},"4",{"type":43,"value":934}," invalid API response.",{"type":38,"tag":241,"props":936,"children":937},{},[938,944,946,952],{"type":38,"tag":54,"props":939,"children":941},{"className":940},[],[942],{"type":43,"value":943},"PWC_API_URL",{"type":43,"value":945}," may select another compatible v1 endpoint. The default is\n",{"type":38,"tag":54,"props":947,"children":949},{"className":948},[],[950],{"type":43,"value":951},"https:\u002F\u002Fpaperswithcode.co\u002Fapi\u002Fv1",{"type":43,"value":217},{"type":38,"tag":39,"props":954,"children":955},{},[956],{"type":43,"value":957},"The standalone parser contains no authentication, mutation, ingestion,\npublication, image, CRON, embedding, or infrastructure commands. Do not\nsubstitute repository-maintenance commands into a public or sandbox workflow.",{"type":38,"tag":959,"props":960,"children":961},"style",{},[962],{"type":43,"value":963},"html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"items":965,"total":21},[966],{"slug":4,"name":4,"fn":5,"description":6,"org":967,"tags":968,"stars":21,"repoUrl":22,"updatedAt":23},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[969,970,971],{"name":19,"slug":20,"type":15},{"name":9,"slug":17,"type":15},{"name":13,"slug":14,"type":15},{"items":973,"total":1146},[974,996,1012,1027,1041,1054,1067,1082,1096,1106,1119,1133],{"slug":975,"name":975,"fn":976,"description":977,"org":978,"tags":979,"stars":993,"repoUrl":994,"updatedAt":995},"train-sentence-transformers","train sentence-transformers models","Train or fine-tune sentence-transformers models across `SentenceTransformer` (bi-encoder; dense or static embedding model; for retrieval, similarity, clustering, classification, paraphrase mining, dedup, multimodal), `CrossEncoder` (reranker; pair scoring for two-stage retrieval \u002F pair classification), and `SparseEncoder` (SPLADE, sparse embedding model; for learned-sparse retrieval). Covers loss selection, hard-negative mining, evaluators, distillation, LoRA, Matryoshka, and Hugging Face Hub publishing. Use for any sentence-transformers training task.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[980,983,984,987,990],{"name":981,"slug":982,"type":15},"Deep Learning","deep-learning",{"name":9,"slug":17,"type":15},{"name":985,"slug":986,"type":15},"LLM","llm",{"name":988,"slug":989,"type":15},"Python","python",{"name":991,"slug":992,"type":15},"Search","search",18914,"https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Fsentence-transformers","2026-05-08T05:09:16.820066",{"slug":997,"name":997,"fn":998,"description":999,"org":1000,"tags":1001,"stars":1009,"repoUrl":1010,"updatedAt":1011},"trl-training","train and fine-tune LLMs with TRL","Train and fine-tune transformer language models using TRL (Transformers Reinforcement Learning). Supports SFT, DPO, GRPO, KTO, RLOO and Reward Model training via CLI commands.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1002,1005,1006,1007,1008],{"name":1003,"slug":1004,"type":15},"AI Infrastructure","ai-infrastructure",{"name":981,"slug":982,"type":15},{"name":9,"slug":17,"type":15},{"name":985,"slug":986,"type":15},{"name":988,"slug":989,"type":15},18850,"https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Ftrl","2026-04-06T18:25:32.746828",{"slug":1013,"name":1013,"fn":1014,"description":1015,"org":1016,"tags":1017,"stars":1024,"repoUrl":1025,"updatedAt":1026},"hf-cli","manage Hugging Face Hub resources via CLI","Hugging Face Hub CLI (`hf`) for downloading, uploading, and managing models, datasets, spaces, buckets, repos, papers, jobs, and more on the Hugging Face Hub. Use when: handling authentication; managing local cache; managing Hugging Face Buckets; running or scheduling jobs on Hugging Face infrastructure; managing Hugging Face repos; discussions and pull requests; browsing models, datasets and spaces; reading, searching, or browsing academic papers; managing collections; querying datasets; configuring spaces; setting up webhooks; or deploying and managing HF Inference Endpoints. Make sure to use this skill whenever the user mentions 'hf', 'huggingface', 'Hugging Face', 'huggingface-cli', or 'hugging face cli', or wants to do anything related to the Hugging Face ecosystem and to AI and ML in general. Also use for cloud storage needs like training checkpoints, data pipelines, or agent traces. Use even if the user doesn't explicitly ask for a CLI command. Replaces the deprecated `huggingface-cli`.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1018,1019,1022,1023],{"name":19,"slug":20,"type":15},{"name":1020,"slug":1021,"type":15},"Datasets","datasets",{"name":9,"slug":17,"type":15},{"name":985,"slug":986,"type":15},10861,"https:\u002F\u002Fgithub.com\u002Fhuggingface\u002Fskills","2026-04-06T18:25:34.020855",{"slug":1028,"name":1028,"fn":1029,"description":1030,"org":1031,"tags":1032,"stars":1024,"repoUrl":1025,"updatedAt":1040},"hf-cloud-aws-context-discovery","discover local AWS environment context","Discover the user's local AWS context (active profile, region, account ID, caller identity) at the start of any AWS task. Use this skill before any other AWS work — deploying to SageMaker, creating resources, calling AWS APIs, or anything that touches an AWS account. Use it especially when the user has not specified a region or profile explicitly, when they say things like \"use my AWS account\", \"deploy to AWS\", \"use my profile\", or when about to make any AWS CLI or SDK call. Never guess the region or account ID — always use this skill to read it from the local configuration first.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1033,1036,1037],{"name":1034,"slug":1035,"type":15},"AWS","aws",{"name":19,"slug":20,"type":15},{"name":1038,"slug":1039,"type":15},"Configuration","configuration","2026-07-08T05:55:33.716099",{"slug":1042,"name":1042,"fn":1043,"description":1044,"org":1045,"tags":1046,"stars":1024,"repoUrl":1025,"updatedAt":1053},"hf-cloud-python-env-setup","set up Python environments for AWS","Set up an isolated Python environment for SageMaker \u002F AWS work, with the right Python version and current boto3. Use this skill whenever Python code will be executed for a SageMaker deployment, training job, or any AWS automation — including when about to run `pip install`, when about to invoke `boto3`, when creating or activating a virtualenv, or when the user asks to \"set up the environment\". Never use system Python and never `pip install` into it. Always isolate. This skill prevents the most common failure modes: wrong Python version, dependency conflicts, and stale SDKs.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1047,1048,1049,1052],{"name":1003,"slug":1004,"type":15},{"name":1034,"slug":1035,"type":15},{"name":1050,"slug":1051,"type":15},"Engineering","engineering",{"name":988,"slug":989,"type":15},"2026-07-08T05:55:32.505017",{"slug":1055,"name":1055,"fn":1056,"description":1057,"org":1058,"tags":1059,"stars":1024,"repoUrl":1025,"updatedAt":1066},"hf-cloud-sagemaker-deployment-planner","plan model deployments to Amazon SageMaker","Plan and coordinate the deployment of a model to Amazon SageMaker AI. Use this skill whenever the user wants to deploy, host, serve, or expose a model on SageMaker or AWS — including phrases like \"deploy a model\", \"host this LLM on AWS\", \"serve this embedding model\", \"deploy a reranker\", \"deploy a text-to-image \u002F diffusion model\", \"host this for async inference\", \"create an endpoint\", \"serve my fine-tuned model\", or any request that involves making a model available for inference on AWS. Use this even when the user is vague (e.g. \"I just want to get this running on AWS, you figure it out\"). Works for text-generation LLMs, embedding models, rerankers, classifiers, text-to-image \u002F diffusion models — picks the right serving stack and chooses between real-time and async inference. This is the entry-point skill for SageMaker deployment work — it asks clarifying questions, picks a deployment pathway, and coordinates the other deployment skills.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1060,1061,1062,1065],{"name":1003,"slug":1004,"type":15},{"name":1034,"slug":1035,"type":15},{"name":1063,"slug":1064,"type":15},"Deployment","deployment",{"name":9,"slug":17,"type":15},"2026-07-08T05:55:37.387689",{"slug":1068,"name":1068,"fn":1069,"description":1070,"org":1071,"tags":1072,"stars":1024,"repoUrl":1025,"updatedAt":1081},"hf-cloud-sagemaker-iam-preflight","configure SageMaker IAM roles","Ensure a usable SageMaker execution role exists before deploying or training. Use this skill whenever about to create a SageMaker endpoint, model, training job, or any resource that requires an execution role. Use it especially when the user has not provided a role ARN explicitly, when scripts are about to call `iam:CreateRole`, or when an AccessDenied error mentions an IAM action. Never blindly call `iam:CreateRole` — always check for existing roles first. This skill prevents the most common SageMaker deployment failure: trying to create IAM resources from an SSO principal that has no IAM write permissions.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1073,1074,1075,1078],{"name":1003,"slug":1004,"type":15},{"name":1034,"slug":1035,"type":15},{"name":1076,"slug":1077,"type":15},"Permissions","permissions",{"name":1079,"slug":1080,"type":15},"Security","security","2026-07-08T05:55:34.948771",{"slug":1083,"name":1083,"fn":1084,"description":1085,"org":1086,"tags":1087,"stars":1024,"repoUrl":1025,"updatedAt":1095},"hf-cloud-sagemaker-production-defaults","create production-ready SageMaker endpoints","Create a SageMaker endpoint (real-time or async) with autoscaling, CloudWatch alarms, and tagging enabled by default. Use this skill whenever about to create a SageMaker endpoint, write deployment code that calls `create_endpoint`, or finalize a deployment after the image URI and IAM role are known. Provides deploy.py for real-time endpoints and deploy_async.py for async endpoints (with genuine scale-to-zero support). This is the last step in the SageMaker deployment workflow. Never generate a bare `create_endpoint` call without these defaults — endpoints without autoscaling or alarms are demos, not deployments.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1088,1089,1090,1091,1092],{"name":1003,"slug":1004,"type":15},{"name":1034,"slug":1035,"type":15},{"name":1063,"slug":1064,"type":15},{"name":9,"slug":17,"type":15},{"name":1093,"slug":1094,"type":15},"Monitoring","monitoring","2026-07-08T05:55:38.664702",{"slug":1097,"name":1097,"fn":1098,"description":1099,"org":1100,"tags":1101,"stars":1024,"repoUrl":1025,"updatedAt":1105},"hf-cloud-serving-image-selection","select SageMaker serving containers","Pick the right serving container for a SageMaker model deployment and find its current image URI. Use this skill whenever about to deploy a model to a SageMaker endpoint and an image URI needs to be chosen — including when the user says \"deploy this LLM\", \"host this HuggingFace model\", \"serve this fine-tuned model\", \"deploy this embedding model\", \"host a reranker\", \"serve a sentence-transformers model\", or when about to hardcode any container URI in deployment code. HuggingFace-curated Deep Learning Containers are ALWAYS preferred: HuggingFace vLLM (LLMs and generative rerankers), HuggingFace vLLM-Omni (multimodal), TEI (embeddings\u002Fcross-encoder rerankers), HF Inference Toolkit (other transformers). Generic images (AWS vLLM, DJL-LMI, SGLang) are used only when no HuggingFace image is compatible — never merely because they carry a newer version. Never hardcode a container URI from memory and never default to TGI. Prevents stale-image failures and wrong-region URIs.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1102,1103,1104],{"name":1003,"slug":1004,"type":15},{"name":1034,"slug":1035,"type":15},{"name":1063,"slug":1064,"type":15},"2026-07-08T05:55:36.173465",{"slug":1107,"name":1107,"fn":1108,"description":1109,"org":1110,"tags":1111,"stars":1024,"repoUrl":1025,"updatedAt":1118},"hf-mcp","access Hugging Face Hub via MCP","Use Hugging Face Hub via MCP server tools. Search models, datasets, Spaces, papers. Get repo details, fetch documentation, run compute jobs, and use Gradio Spaces as AI tools. Available when connected to the HF MCP server.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1112,1113,1114,1115],{"name":1020,"slug":1021,"type":15},{"name":9,"slug":17,"type":15},{"name":985,"slug":986,"type":15},{"name":1116,"slug":1117,"type":15},"MCP","mcp","2026-04-06T18:25:50.364185",{"slug":1120,"name":1120,"fn":1121,"description":1122,"org":1123,"tags":1124,"stars":1024,"repoUrl":1025,"updatedAt":1132},"hf-mem","estimate memory for Hugging Face models","Hugging Face CLI to estimate the required memory to load Safetensors or GGUF model weights for inference from the Hugging Face Hub",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1125,1126,1127,1128,1129],{"name":1003,"slug":1004,"type":15},{"name":19,"slug":20,"type":15},{"name":9,"slug":17,"type":15},{"name":985,"slug":986,"type":15},{"name":1130,"slug":1131,"type":15},"Performance","performance","2026-06-13T07:23:57.101435",{"slug":1134,"name":1134,"fn":1135,"description":1136,"org":1137,"tags":1138,"stars":1024,"repoUrl":1025,"updatedAt":1145},"huggingface-best","find and compare Hugging Face models","Use when the user asks about finding the best, top, or recommended model for a task, wants to know what AI model to use, or wants to compare models by benchmark scores. Triggers on: \"best model for X\", \"what model should I use for\", \"top models for [task]\", \"which model runs on my laptop\u002Fmachine\u002Fdevice\", \"recommend a model for\", \"what LLM should I use for\", \"compare models for\", \"what's state of the art for\", or any question about choosing an AI model for a specific use case. Always use this skill when the user wants model recommendations or comparisons, even if they don't explicitly mention HuggingFace or benchmarks.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1139,1142,1143,1144],{"name":1140,"slug":1141,"type":15},"Analytics","analytics",{"name":9,"slug":17,"type":15},{"name":985,"slug":986,"type":15},{"name":13,"slug":14,"type":15},"2026-04-24T05:09:45.870658",37]