[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"extension-skill-huggingface-huggingface-community-evals-en":3,"guides-for-huggingface-huggingface-community-evals":704,"similar-k1726vs7xfdjf1be08gkbz6v1s86mxk2-en":705},{"_creationTime":4,"_id":5,"children":6,"community":7,"display":9,"evaluation":15,"identity":213,"isFallback":196,"parentExtension":217,"providers":251,"relations":255,"repo":256,"tags":702,"workflow":703},1778690773482.487,"k1726vs7xfdjf1be08gkbz6v1s86mxk2",[],{"reviewCount":8},0,{"description":10,"installMethods":11,"name":13,"sourceUrl":14},"Run evaluations for Hugging Face Hub models using inspect-ai and lighteval on local hardware. Use for backend selection, local GPU evals, and choosing between vLLM / Transformers / accelerate. Not for HF Jobs orchestration, model-card PRs, .eval_results publication, or community-evals automation.",{"claudeCode":12},"huggingface/skills","huggingface-community-evals","https://github.com/huggingface/skills",{"_creationTime":16,"_id":17,"extensionId":5,"locale":18,"result":19,"trustSignals":194,"workflow":211},1778691261342.1392,"kn7e9f08x025q6zac9qd0xnbw986nf5h","en",{"checks":20,"evaluatedAt":160,"extensionSummary":161,"features":162,"nonGoals":168,"promptVersionExtension":173,"promptVersionScoring":174,"purpose":175,"rationale":176,"score":177,"summary":178,"tags":179,"targetMarket":187,"tier":188,"useCases":189},[21,26,29,32,36,39,43,47,50,53,57,61,64,68,71,74,77,80,83,86,90,94,98,102,106,109,113,116,120,123,126,129,132,135,138,142,146,149,153,157],{"category":22,"check":23,"severity":24,"summary":25},"Practical Utility","Problem relevance","pass","The description clearly states the problem of running local Hugging Face Hub model evaluations using specific tools and local hardware, providing concrete use cases and exclusions.",{"category":22,"check":27,"severity":24,"summary":28},"Unique selling proposition","The skill offers value beyond basic prompting by providing structured workflows and specific configurations for `inspect-ai` and `lighteval` with local inference backends, addressing the complexity of choosing and configuring these tools.",{"category":22,"check":30,"severity":24,"summary":31},"Production readiness","The skill provides ready-to-use scripts for local evaluations, including setup instructions and troubleshooting, indicating it can be used in a real workflow.",{"category":33,"check":34,"severity":24,"summary":35},"Scope","Single responsibility principle","The skill focuses solely on running local Hugging Face model evaluations using `inspect-ai` and `lighteval`, with clear boundaries defined for what it does not cover.",{"category":33,"check":37,"severity":24,"summary":38},"Description quality","The description is concise, accurate, and clearly communicates the skill's purpose, capabilities, and limitations.",{"category":40,"check":41,"severity":24,"summary":42},"Invocation","Scoped tools","The skill utilizes well-defined scripts (`inspect_eval_uv.py`, `inspect_vllm_uv.py`, `lighteval_vllm_uv.py`) for specific evaluation tasks, rather than a single generalist command.",{"category":44,"check":45,"severity":24,"summary":46},"Documentation","Configuration & parameter reference","The `SKILL.md` and script argument parsers clearly document all options and parameters, including defaults, and specify prerequisites.",{"category":33,"check":48,"severity":24,"summary":49},"Tool naming","The scripts are named descriptively (`inspect_eval_uv.py`, `lighteval_vllm_uv.py`) and their purpose is clear from the context.",{"category":33,"check":51,"severity":24,"summary":52},"Minimal I/O surface","The script arguments are specific to the evaluation task, and the expected output is the evaluation results without unnecessary diagnostic dumps.",{"category":54,"check":55,"severity":24,"summary":56},"License","License usability","The extension uses the Apache-2.0 license, which is a permissive open-source license, as indicated by the bundled LICENSE file and trust signals.",{"category":58,"check":59,"severity":24,"summary":60},"Maintenance","Commit recency","The last commit was on 2026-05-12, which is recent, indicating active maintenance.",{"category":58,"check":62,"severity":24,"summary":63},"Dependency Management","Dependencies are managed via `pyproject.toml` and `poetry.lock` (inferred from uv usage and dependency list in script headers), indicating proper management.",{"category":65,"check":66,"severity":24,"summary":67},"Security","Secret Management","The skill handles secrets by reading `HF_TOKEN` from environment variables and setting `HUGGING_FACE_HUB_TOKEN`, without echoing them to stdout.",{"category":65,"check":69,"severity":24,"summary":70},"Injection","The skill executes specific, well-defined Python scripts with parsed arguments, and does not load or execute arbitrary third-party data as instructions.",{"category":65,"check":72,"severity":24,"summary":73},"Transitive Supply-Chain Grenades","The skill relies on bundled Python scripts and standard dependencies managed via `uv`, with no runtime downloads of code or data not included in the bundle.",{"category":65,"check":75,"severity":24,"summary":76},"Sandbox Isolation","The scripts are executed within a controlled environment using `uv run`, and do not appear to modify files outside the project directory or assume specific user-specific paths.",{"category":65,"check":78,"severity":24,"summary":79},"Sandbox escape primitives","No detached process spawns or deny-retry loops were detected in the provided script code.",{"category":65,"check":81,"severity":24,"summary":82},"Data Exfiltration","The skill does not instruct the agent to read or submit confidential data to a third party; Hugging Face token is used for authentication only.",{"category":65,"check":84,"severity":24,"summary":85},"Hidden Text Tricks","The bundled files and scripts are free of hidden steering tricks, invisible characters, or obfuscation designed to mislead the model.",{"category":87,"check":88,"severity":24,"summary":89},"Hooks","Opaque code execution","The provided scripts are plain Python code and do not use obfuscation techniques like base64 encoding, eval, or runtime script fetching.",{"category":91,"check":92,"severity":24,"summary":93},"Portability","Structural Assumption","The scripts are designed to run within the skill's directory structure and do not make assumptions about external project layouts.",{"category":95,"check":96,"severity":24,"summary":97},"Trust","Issues Attention","Recent activity shows 4 issues opened and 6 closed in the last 90 days, indicating active maintenance and responsiveness.",{"category":99,"check":100,"severity":24,"summary":101},"Versioning","Release Management","The `requires-python` and `dependencies` directives in the script headers, along with the clear use of `uv` for environment management, imply versioned dependencies and a manageable release process.",{"category":103,"check":104,"severity":24,"summary":105},"Execution","Validation","The Python scripts use `argparse` for argument parsing, providing a structured way to handle inputs, and the types are defined, implying validation.",{"category":65,"check":107,"severity":24,"summary":108},"Unguarded Destructive Operations","The skill is primarily for running evaluations, which are read-only operations, and does not involve destructive primitives.",{"category":110,"check":111,"severity":24,"summary":112},"Code Execution","Error Handling","The scripts use `subprocess.run(check=True)` and `try...except subprocess.CalledProcessError` blocks, ensuring that script execution errors are caught and reported.",{"category":110,"check":114,"severity":24,"summary":115},"Logging","The scripts include print statements for progress and error reporting, and the `inspect` tool has a `--log-level` argument, indicating basic logging capabilities.",{"category":117,"check":118,"severity":24,"summary":119},"Compliance","GDPR","The skill operates on model evaluations and Hugging Face tokens for authentication, without directly processing personal user data beyond authentication credentials.",{"category":117,"check":121,"severity":24,"summary":122},"Target market","The skill's functionality is global, focused on model evaluation tools, and does not exhibit any regional or jurisdictional logic.",{"category":91,"check":124,"severity":24,"summary":125},"Runtime stability","The scripts specify Python 3.10+ requirements and common dependencies, suggesting good portability across standard Python environments.",{"category":44,"check":127,"severity":24,"summary":128},"README","The README provides a comprehensive overview of the Hugging Face Skills repository, including installation, usage, and contribution guidelines.",{"category":33,"check":130,"severity":24,"summary":131},"Tool surface size","The extension primarily consists of three main Python scripts for different evaluation scenarios, which is a reasonable number.",{"category":40,"check":133,"severity":24,"summary":134},"Overlapping near-synonym tools","The skill uses distinct scripts for different evaluation frameworks (`inspect-ai`, `lighteval`) and backends (`vllm`, `hf`, `accelerate`), avoiding naming conflicts.",{"category":44,"check":136,"severity":24,"summary":137},"Phantom features","All features advertised in the SKILL.md and README (local evaluations, backend selection) are directly implemented by the provided Python scripts.",{"category":139,"check":140,"severity":24,"summary":141},"Install","Installation instruction","The README provides clear installation instructions for various agents (Claude Code, Codex, Gemini CLI, Cursor) and the SKILL.md outlines prerequisites.",{"category":143,"check":144,"severity":24,"summary":145},"Errors","Actionable error messages","The scripts provide informative error messages when `subprocess.run` fails, indicating the exit code and directing users to investigate the failure.",{"category":103,"check":147,"severity":24,"summary":148},"Pinned dependencies","Dependencies are declared with version constraints in script headers (e.g., `lighteval>=0.6.0`, `vllm>=0.4.0`), indicating pinned versions for reproducibility.",{"category":33,"check":150,"severity":151,"summary":152},"Dry-run preview","not_applicable","This skill is for running evaluations, which are inherently read-only operations and do not involve state changes or outbound data that would require a dry-run preview.",{"category":154,"check":155,"severity":24,"summary":156},"Protocol","Idempotent retry & timeouts","The core operations are evaluations, which are typically idempotent. The scripts use `subprocess.run(check=True)` which implies some level of error handling for execution, and `inspect-ai` itself likely has timeouts. No external API calls are made that would require explicit retry logic here.",{"category":117,"check":158,"severity":24,"summary":159},"Telemetry opt-in","There is no indication of telemetry being collected or emitted by this skill; all operations are local evaluations.",1778691260607,"This skill provides Python scripts to run local evaluations against Hugging Face Hub models using `inspect-ai` and `lighteval`. It supports `vLLM`, Hugging Face Transformers, and `accelerate` backends, offering configuration for model selection, task execution, and GPU utilization.",[163,164,165,166,167],"Run `inspect-ai` evaluations locally","Run `lighteval` evaluations locally","Support for `vLLM`, Transformers, and `accelerate` backends","Configuration for model selection and task execution","Guidance on local GPU requirements and troubleshooting",[169,170,171,172],"Orchestrating Hugging Face Jobs for remote evaluations.","Editing model cards, `model-index` files, or publishing `.eval_results`.","Automating community evaluations or PRs.","Directly managing Hugging Face Hub resources beyond model evaluation.","3.0.0","4.4.0","To enable users to run local, GPU-accelerated evaluations of Hugging Face Hub models for backend selection and performance comparisons.","Excellent quality and implementation. No critical or warning findings.",98,"A high-quality skill for running local Hugging Face model evaluations using inspect-ai and lighteval.",[180,181,182,183,184,185,186],"huggingface","evaluation","llm","inspect-ai","lighteval","gpu","vllm","global","verified",[190,191,192,193],"Selecting the best inference backend (vLLM, Transformers, accelerate) for local model evaluations.","Performing smoke tests and full evaluations on Hugging Face Hub models using local hardware.","Comparing model performance across different tasks and backends without relying on cloud services.","Troubleshooting local GPU evaluation setup issues.",{"codeQuality":195,"collectedAt":197,"documentation":198,"maintenance":201,"security":207,"testCoverage":209},{"hasLockfile":196},false,1778691241540,{"descriptionLength":199,"readmeSize":200},297,9821,{"closedIssues90d":202,"forks":203,"hasChangelog":196,"openIssues90d":204,"pushedAt":205,"stars":206},6,663,4,1778593131000,10482,{"hasNpmPackage":196,"license":208,"smitheryVerified":196},"Apache-2.0",{"hasCi":210,"hasTests":196},true,{"updatedAt":212},1778691261342,{"basePath":214,"githubOwner":180,"githubRepo":215,"locale":18,"slug":13,"type":216},"skills/huggingface-community-evals","skills","skill",{"_creationTime":218,"_id":219,"community":220,"display":221,"identity":226,"parentExtension":229,"providers":230,"relations":245,"tags":247,"workflow":248},1778690773482.486,"k175g1spb5757qt4tnj9cktcn986mshy",{"reviewCount":8},{"description":222,"installMethods":223,"name":225,"sourceUrl":14},"Agent Skills for AI/ML tasks including dataset creation, model training, evaluation, and research paper publishing on Hugging Face Hub",{"claudeCode":224},"huggingface-skills","Hugging Face Skills",{"basePath":227,"githubOwner":180,"githubRepo":215,"locale":18,"slug":215,"type":228},"","plugin",null,{"evaluate":231,"extract":240},{"promptVersionExtension":173,"promptVersionScoring":174,"score":177,"tags":232,"targetMarket":187,"tier":188},[180,233,234,235,236,237,238,239],"ai","ml","datasets","models","training","cli","python",{"commitSha":241,"license":208,"plugin":242},"HEAD",{"mcpCount":8,"provider":243,"skillCount":244},"classify",14,{"repoId":246},"kd72xwt5xnc0ktc4p7smzfcp3986m959",[233,238,235,180,234,236,239,237],{"evaluatedAt":249,"extractAt":250,"updatedAt":249},1778691185872,1778690773482,{"evaluate":252,"extract":254},{"promptVersionExtension":173,"promptVersionScoring":174,"score":177,"tags":253,"targetMarket":187,"tier":188},[180,181,182,183,184,185,186],{"commitSha":241},{"parentExtensionId":219,"repoId":246},{"_creationTime":257,"_id":246,"identity":258,"providers":259,"workflow":698},1778689536128.5474,{"githubOwner":180,"githubRepo":215,"sourceUrl":14},{"classify":260,"discover":691,"github":694},{"commitSha":241,"extensions":261},[262,276,285,293,301,309,315,323,331,339,347,355,363,371,379,387,430,438,444,450,466,472,479,521,532,551,557,577,589,613,671],{"basePath":227,"description":222,"displayName":224,"installMethods":263,"rationale":264,"selectedPaths":265,"source":274,"sourceLanguage":18,"type":275},{"claudeCode":12},"marketplace.json at .claude-plugin/marketplace.json",[266,269,271],{"path":267,"priority":268},".claude-plugin/marketplace.json","mandatory",{"path":270,"priority":268},"README.md",{"path":272,"priority":273},"LICENSE","high","rule","marketplace",{"basePath":277,"description":278,"displayName":279,"installMethods":280,"rationale":281,"selectedPaths":282,"source":274,"sourceLanguage":18,"type":228},"skills/huggingface-llm-trainer","Train or fine-tune language models using TRL on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes hardware selection, cost estimation, Trackio monitoring, and Hub persistence.","huggingface-llm-trainer",{"claudeCode":279},"inline plugin source from marketplace.json at skills/huggingface-llm-trainer",[283],{"path":284,"priority":273},"SKILL.md",{"basePath":286,"description":287,"displayName":288,"installMethods":289,"rationale":290,"selectedPaths":291,"source":274,"sourceLanguage":18,"type":228},"skills/huggingface-local-models","Use to select models to run locally with llama.cpp and GGUF on CPU, Mac Metal, CUDA, or ROCm. Covers finding GGUFs, quant selection, running servers, exact GGUF file lookup, conversion, and OpenAI-compatible local serving.","huggingface-local-models",{"claudeCode":288},"inline plugin source from marketplace.json at skills/huggingface-local-models",[292],{"path":284,"priority":273},{"basePath":294,"description":295,"displayName":296,"installMethods":297,"rationale":298,"selectedPaths":299,"source":274,"sourceLanguage":18,"type":228},"skills/huggingface-paper-publisher","Publish and manage research papers on Hugging Face Hub. Supports creating paper pages, linking papers to models/datasets, claiming authorship, and generating professional markdown-based research articles.","huggingface-paper-publisher",{"claudeCode":296},"inline plugin source from marketplace.json at skills/huggingface-paper-publisher",[300],{"path":284,"priority":273},{"basePath":302,"description":303,"displayName":304,"installMethods":305,"rationale":306,"selectedPaths":307,"source":274,"sourceLanguage":18,"type":228},"skills/huggingface-papers","Look up and read Hugging Face paper pages in markdown, and use the papers API for structured metadata like authors, linked models, datasets, Spaces, and media URLs when needed.","huggingface-papers",{"claudeCode":304},"inline plugin source from marketplace.json at skills/huggingface-papers",[308],{"path":284,"priority":273},{"basePath":214,"description":310,"displayName":13,"installMethods":311,"rationale":312,"selectedPaths":313,"source":274,"sourceLanguage":18,"type":228},"Add and manage evaluation results in Hugging Face model cards. Supports extracting eval tables from README content, importing scores from Artificial Analysis API, and running custom evaluations with vLLM/lighteval.",{"claudeCode":13},"inline plugin source from marketplace.json at skills/huggingface-community-evals",[314],{"path":284,"priority":273},{"basePath":316,"description":317,"displayName":318,"installMethods":319,"rationale":320,"selectedPaths":321,"source":274,"sourceLanguage":18,"type":228},"skills/huggingface-best","Find the best AI model for any task by querying Hugging Face leaderboards and benchmarks. Recommends top models based on task type, hardware constraints, and benchmark scores.","huggingface-best",{"claudeCode":318},"inline plugin source from marketplace.json at skills/huggingface-best",[322],{"path":284,"priority":273},{"basePath":324,"description":325,"displayName":326,"installMethods":327,"rationale":328,"selectedPaths":329,"source":274,"sourceLanguage":18,"type":228},"skills/hf-cli","Execute Hugging Face Hub operations using the hf CLI. Download models/datasets, upload files, manage repos, and run cloud compute jobs.","hf-cli",{"claudeCode":326},"inline plugin source from marketplace.json at skills/hf-cli",[330],{"path":284,"priority":273},{"basePath":332,"description":333,"displayName":334,"installMethods":335,"rationale":336,"selectedPaths":337,"source":274,"sourceLanguage":18,"type":228},"skills/huggingface-trackio","Track and visualize ML training experiments with Trackio. Log metrics via Python API and retrieve them via CLI. Supports real-time dashboards synced to HF Spaces.","huggingface-trackio",{"claudeCode":334},"inline plugin source from marketplace.json at skills/huggingface-trackio",[338],{"path":284,"priority":273},{"basePath":340,"description":341,"displayName":342,"installMethods":343,"rationale":344,"selectedPaths":345,"source":274,"sourceLanguage":18,"type":228},"skills/huggingface-datasets","Explore, query, and extract data from any Hugging Face dataset using the Dataset Viewer REST API and npx tooling. Zero Python dependencies — covers split/config discovery, row pagination, text search, filtering, SQL via parquetlens, and dataset upload via CLI.","huggingface-datasets",{"claudeCode":342},"inline plugin source from marketplace.json at skills/huggingface-datasets",[346],{"path":284,"priority":273},{"basePath":348,"description":349,"displayName":350,"installMethods":351,"rationale":352,"selectedPaths":353,"source":274,"sourceLanguage":18,"type":228},"skills/huggingface-tool-builder","Build reusable scripts for Hugging Face Hub and API workflows. Useful for chaining API calls, enriching Hub metadata, or automating repeated tasks.","huggingface-tool-builder",{"claudeCode":350},"inline plugin source from marketplace.json at skills/huggingface-tool-builder",[354],{"path":284,"priority":273},{"basePath":356,"description":357,"displayName":358,"installMethods":359,"rationale":360,"selectedPaths":361,"source":274,"sourceLanguage":18,"type":228},"skills/huggingface-gradio","Build Gradio web UIs and demos in Python. Use when creating or editing Gradio apps, components, event listeners, layouts, or chatbots.","huggingface-gradio",{"claudeCode":358},"inline plugin source from marketplace.json at skills/huggingface-gradio",[362],{"path":284,"priority":273},{"basePath":364,"description":365,"displayName":366,"installMethods":367,"rationale":368,"selectedPaths":369,"source":274,"sourceLanguage":18,"type":228},"skills/transformers-js","Run state-of-the-art machine learning models directly in JavaScript/TypeScript for NLP, computer vision, audio processing, and multimodal tasks. Works in Node.js and browsers with WebGPU/WASM using Hugging Face models.","transformers-js",{"claudeCode":366},"inline plugin source from marketplace.json at skills/transformers-js",[370],{"path":284,"priority":273},{"basePath":372,"description":373,"displayName":374,"installMethods":375,"rationale":376,"selectedPaths":377,"source":274,"sourceLanguage":18,"type":228},"skills/huggingface-vision-trainer","Train and fine-tune object detection models (RTDETRv2, YOLOS, DETR and others) and image classification models (timm and transformers models — MobileNetV3, MobileViT, ResNet, ViT/DINOv3) using Transformers Trainer API on Hugging Face Jobs infrastructure or locally. Includes COCO dataset format support, Albumentations augmentation, mAP/mAR metrics, trackio tracking, hardware selection, and Hub persistence.","huggingface-vision-trainer",{"claudeCode":374},"inline plugin source from marketplace.json at skills/huggingface-vision-trainer",[378],{"path":284,"priority":273},{"basePath":380,"description":381,"displayName":382,"installMethods":383,"rationale":384,"selectedPaths":385,"source":274,"sourceLanguage":18,"type":228},"skills/train-sentence-transformers","Train or fine-tune sentence-transformers models across all three architectures: SentenceTransformer (bi-encoder embeddings), CrossEncoder (rerankers), and SparseEncoder (SPLADE). Covers loss selection, hard-negative mining, evaluators, distillation, LoRA, Matryoshka, and Hugging Face Hub publishing.","train-sentence-transformers",{"claudeCode":382},"inline plugin source from marketplace.json at skills/train-sentence-transformers",[386],{"path":284,"priority":273},{"basePath":227,"description":222,"displayName":224,"installMethods":388,"license":208,"rationale":389,"selectedPaths":390,"source":274,"sourceLanguage":18,"type":228},{"claudeCode":224},"plugin manifest at .claude-plugin/plugin.json",[391,393,394,395,398,400,402,404,406,408,410,412,414,416,418,420,422,424,426,428],{"path":392,"priority":268},".claude-plugin/plugin.json",{"path":270,"priority":268},{"path":272,"priority":273},{"path":396,"priority":397},"skills/hf-cli/SKILL.md","medium",{"path":399,"priority":397},"skills/huggingface-best/SKILL.md",{"path":401,"priority":397},"skills/huggingface-community-evals/SKILL.md",{"path":403,"priority":397},"skills/huggingface-datasets/SKILL.md",{"path":405,"priority":397},"skills/huggingface-gradio/SKILL.md",{"path":407,"priority":397},"skills/huggingface-llm-trainer/SKILL.md",{"path":409,"priority":397},"skills/huggingface-local-models/SKILL.md",{"path":411,"priority":397},"skills/huggingface-paper-publisher/SKILL.md",{"path":413,"priority":397},"skills/huggingface-papers/SKILL.md",{"path":415,"priority":397},"skills/huggingface-tool-builder/SKILL.md",{"path":417,"priority":397},"skills/huggingface-trackio/SKILL.md",{"path":419,"priority":397},"skills/huggingface-vision-trainer/SKILL.md",{"path":421,"priority":397},"skills/train-sentence-transformers/SKILL.md",{"path":423,"priority":397},"skills/transformers-js/SKILL.md",{"path":425,"priority":268},".mcp.json",{"path":427,"priority":273},"agents/AGENTS.md",{"path":429,"priority":273},".cursor-plugin/plugin.json",{"basePath":431,"description":432,"displayName":433,"installMethods":434,"rationale":435,"selectedPaths":436,"source":274,"sourceLanguage":18,"type":216},"hf-mcp/skills/hf-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.","hf-mcp",{"claudeCode":12},"SKILL.md frontmatter at hf-mcp/skills/hf-mcp/SKILL.md",[437],{"path":284,"priority":268},{"basePath":324,"description":439,"displayName":326,"installMethods":440,"rationale":441,"selectedPaths":442,"source":274,"sourceLanguage":18,"type":216},"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`.",{"claudeCode":12},"SKILL.md frontmatter at skills/hf-cli/SKILL.md",[443],{"path":284,"priority":268},{"basePath":316,"description":445,"displayName":318,"installMethods":446,"rationale":447,"selectedPaths":448,"source":274,"sourceLanguage":18,"type":216},"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/machine/device\", \"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",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-best/SKILL.md",[449],{"path":284,"priority":268},{"basePath":214,"description":10,"displayName":13,"installMethods":451,"rationale":452,"selectedPaths":453,"source":274,"sourceLanguage":18,"type":216},{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-community-evals/SKILL.md",[454,455,458,460,462,464],{"path":284,"priority":268},{"path":456,"priority":457},"examples/.env.example","low",{"path":459,"priority":457},"examples/USAGE_EXAMPLES.md",{"path":461,"priority":457},"scripts/inspect_eval_uv.py",{"path":463,"priority":457},"scripts/inspect_vllm_uv.py",{"path":465,"priority":457},"scripts/lighteval_vllm_uv.py",{"basePath":340,"description":467,"displayName":342,"installMethods":468,"rationale":469,"selectedPaths":470,"source":274,"sourceLanguage":18,"type":216},"Use this skill for Hugging Face Dataset Viewer API workflows that fetch subset/split metadata, paginate rows, search text, apply filters, download parquet URLs, and read size or statistics.\r",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-datasets/SKILL.md",[471],{"path":284,"priority":268},{"basePath":356,"description":357,"displayName":358,"installMethods":473,"rationale":474,"selectedPaths":475,"source":274,"sourceLanguage":18,"type":216},{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-gradio/SKILL.md",[476,477],{"path":284,"priority":268},{"path":478,"priority":397},"examples.md",{"basePath":277,"description":480,"displayName":279,"installMethods":481,"rationale":482,"selectedPaths":483,"source":274,"sourceLanguage":18,"type":216},"Train or fine-tune language and vision models using TRL (Transformer Reinforcement Learning) or Unsloth with Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV scripts with PEP 723 format, dataset preparation and validation, hardware selection, cost estimation, Trackio monitoring, Hub authentication, model selection/leaderboards and model persistence. Use for tasks involving cloud GPU training, GGUF conversion, or when users mention training on Hugging Face Jobs without local GPU setup.",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-llm-trainer/SKILL.md",[484,485,487,489,491,493,495,497,499,501,503,505,507,509,511,513,515,517,519],{"path":284,"priority":268},{"path":486,"priority":397},"references/gguf_conversion.md",{"path":488,"priority":397},"references/hardware_guide.md",{"path":490,"priority":397},"references/hub_saving.md",{"path":492,"priority":397},"references/local_training_macos.md",{"path":494,"priority":397},"references/reliability_principles.md",{"path":496,"priority":397},"references/trackio_guide.md",{"path":498,"priority":397},"references/training_methods.md",{"path":500,"priority":397},"references/training_patterns.md",{"path":502,"priority":397},"references/troubleshooting.md",{"path":504,"priority":397},"references/unsloth.md",{"path":506,"priority":457},"scripts/convert_to_gguf.py",{"path":508,"priority":457},"scripts/dataset_inspector.py",{"path":510,"priority":457},"scripts/estimate_cost.py",{"path":512,"priority":457},"scripts/hf_benchmarks.py",{"path":514,"priority":457},"scripts/train_dpo_example.py",{"path":516,"priority":457},"scripts/train_grpo_example.py",{"path":518,"priority":457},"scripts/train_sft_example.py",{"path":520,"priority":457},"scripts/unsloth_sft_example.py",{"basePath":286,"description":287,"displayName":288,"installMethods":522,"rationale":523,"selectedPaths":524,"source":274,"sourceLanguage":18,"type":216},{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-local-models/SKILL.md",[525,526,528,530],{"path":284,"priority":268},{"path":527,"priority":397},"references/hardware.md",{"path":529,"priority":397},"references/hub-discovery.md",{"path":531,"priority":397},"references/quantization.md",{"basePath":294,"description":295,"displayName":296,"installMethods":533,"rationale":534,"selectedPaths":535,"source":274,"sourceLanguage":18,"type":216},{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-paper-publisher/SKILL.md",[536,537,539,541,543,545,547,549],{"path":284,"priority":268},{"path":538,"priority":457},"examples/example_usage.md",{"path":540,"priority":397},"references/quick_reference.md",{"path":542,"priority":457},"scripts/paper_manager.py",{"path":544,"priority":457},"templates/arxiv.md",{"path":546,"priority":457},"templates/ml-report.md",{"path":548,"priority":457},"templates/modern.md",{"path":550,"priority":457},"templates/standard.md",{"basePath":302,"description":552,"displayName":304,"installMethods":553,"rationale":554,"selectedPaths":555,"source":274,"sourceLanguage":18,"type":216},"Look up and read Hugging Face paper pages in markdown, and use the papers API for structured metadata such as authors, linked models/datasets/spaces, Github repo and project page. Use when the user shares a Hugging Face paper page URL, an arXiv URL or ID, or asks to summarize, explain, or analyze an AI research paper.",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-papers/SKILL.md",[556],{"path":284,"priority":268},{"basePath":348,"description":558,"displayName":350,"installMethods":559,"rationale":560,"selectedPaths":561,"source":274,"sourceLanguage":18,"type":216},"Use this skill when the user wants to build tool/scripts or achieve a task where using data from the Hugging Face API would help. This is especially useful when chaining or combining API calls or the task will be repeated/automated. This Skill creates a reusable script to fetch, enrich or process data.",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-tool-builder/SKILL.md",[562,563,565,567,569,571,573,575],{"path":284,"priority":268},{"path":564,"priority":397},"references/baseline_hf_api.py",{"path":566,"priority":397},"references/baseline_hf_api.sh",{"path":568,"priority":397},"references/baseline_hf_api.tsx",{"path":570,"priority":397},"references/find_models_by_paper.sh",{"path":572,"priority":397},"references/hf_enrich_models.sh",{"path":574,"priority":397},"references/hf_model_card_frontmatter.sh",{"path":576,"priority":397},"references/hf_model_papers_auth.sh",{"basePath":332,"description":578,"displayName":334,"installMethods":579,"rationale":580,"selectedPaths":581,"source":274,"sourceLanguage":18,"type":216},"Track and visualize ML training experiments with Trackio. Use when logging metrics during training (Python API), firing alerts for training diagnostics, or retrieving/analyzing logged metrics (CLI). Supports real-time dashboard visualization, alerts with webhooks, HF Space syncing, and JSON output for automation.",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-trackio/SKILL.md",[582,583,585,587],{"path":284,"priority":268},{"path":584,"priority":397},"references/alerts.md",{"path":586,"priority":397},"references/logging_metrics.md",{"path":588,"priority":397},"references/retrieving_metrics.md",{"basePath":372,"description":590,"displayName":374,"installMethods":591,"rationale":592,"selectedPaths":593,"source":274,"sourceLanguage":18,"type":216},"Trains and fine-tunes vision models for object detection (D-FINE, RT-DETR v2, DETR, YOLOS), image classification (timm models — MobileNetV3, MobileViT, ResNet, ViT/DINOv3 — plus any Transformers classifier), and SAM/SAM2 segmentation using Hugging Face Transformers on Hugging Face Jobs cloud GPUs. Covers COCO-format dataset preparation, Albumentations augmentation, mAP/mAR evaluation, accuracy metrics, SAM segmentation with bbox/point prompts, DiceCE loss, hardware selection, cost estimation, Trackio monitoring, and Hub persistence. Use when users mention training object detection, image classification, SAM, SAM2, segmentation, image matting, DETR, D-FINE, RT-DETR, ViT, timm, MobileNet, ResNet, bounding box models, or fine-tuning vision models on Hugging Face Jobs.",{"claudeCode":12},"SKILL.md frontmatter at skills/huggingface-vision-trainer/SKILL.md",[594,595,597,598,600,602,603,605,606,607,609,611],{"path":284,"priority":268},{"path":596,"priority":397},"references/finetune_sam2_trainer.md",{"path":490,"priority":397},{"path":599,"priority":397},"references/image_classification_training_notebook.md",{"path":601,"priority":397},"references/object_detection_training_notebook.md",{"path":494,"priority":397},{"path":604,"priority":397},"references/timm_trainer.md",{"path":508,"priority":457},{"path":510,"priority":457},{"path":608,"priority":457},"scripts/image_classification_training.py",{"path":610,"priority":457},"scripts/object_detection_training.py",{"path":612,"priority":457},"scripts/sam_segmentation_training.py",{"basePath":380,"description":614,"displayName":382,"installMethods":615,"rationale":616,"selectedPaths":617,"source":274,"sourceLanguage":18,"type":216},"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 / 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.",{"claudeCode":12},"SKILL.md frontmatter at skills/train-sentence-transformers/SKILL.md",[618,619,621,623,625,627,629,630,632,634,636,638,640,642,644,645,647,649,651,653,655,657,659,661,663,665,667,669],{"path":284,"priority":268},{"path":620,"priority":397},"references/base_model_selection.md",{"path":622,"priority":397},"references/dataset_formats.md",{"path":624,"priority":397},"references/evaluators_cross_encoder.md",{"path":626,"priority":397},"references/evaluators_sentence_transformer.md",{"path":628,"priority":397},"references/evaluators_sparse_encoder.md",{"path":488,"priority":397},{"path":631,"priority":397},"references/hf_jobs_execution.md",{"path":633,"priority":397},"references/losses_cross_encoder.md",{"path":635,"priority":397},"references/losses_sentence_transformer.md",{"path":637,"priority":397},"references/losses_sparse_encoder.md",{"path":639,"priority":397},"references/model_architectures.md",{"path":641,"priority":397},"references/prompts_and_instructions.md",{"path":643,"priority":397},"references/training_args.md",{"path":502,"priority":397},{"path":646,"priority":457},"scripts/mine_hard_negatives.py",{"path":648,"priority":457},"scripts/train_cross_encoder_distillation_example.py",{"path":650,"priority":457},"scripts/train_cross_encoder_example.py",{"path":652,"priority":457},"scripts/train_cross_encoder_listwise_example.py",{"path":654,"priority":457},"scripts/train_sentence_transformer_distillation_example.py",{"path":656,"priority":457},"scripts/train_sentence_transformer_example.py",{"path":658,"priority":457},"scripts/train_sentence_transformer_make_multilingual_example.py",{"path":660,"priority":457},"scripts/train_sentence_transformer_matryoshka_example.py",{"path":662,"priority":457},"scripts/train_sentence_transformer_multi_dataset_example.py",{"path":664,"priority":457},"scripts/train_sentence_transformer_static_embedding_example.py",{"path":666,"priority":457},"scripts/train_sentence_transformer_with_lora_example.py",{"path":668,"priority":457},"scripts/train_sparse_encoder_distillation_example.py",{"path":670,"priority":457},"scripts/train_sparse_encoder_example.py",{"basePath":364,"description":672,"displayName":366,"installMethods":673,"rationale":674,"selectedPaths":675,"source":274,"sourceLanguage":18,"type":216},"Use Transformers.js to run state-of-the-art machine learning models directly in JavaScript/TypeScript. Supports NLP (text classification, translation, summarization), computer vision (image classification, object detection), audio (speech recognition, audio classification), and multimodal tasks. Works in browsers and server-side runtimes (Node.js, Bun, Deno) with WebGPU/WASM using pre-trained models from Hugging Face Hub.",{"claudeCode":12},"SKILL.md frontmatter at skills/transformers-js/SKILL.md",[676,677,679,681,683,685,687,689],{"path":284,"priority":268},{"path":678,"priority":397},"references/CACHE.md",{"path":680,"priority":397},"references/CONFIGURATION.md",{"path":682,"priority":397},"references/EXAMPLES.md",{"path":684,"priority":397},"references/MODEL_ARCHITECTURES.md",{"path":686,"priority":397},"references/MODEL_REGISTRY.md",{"path":688,"priority":397},"references/PIPELINE_OPTIONS.md",{"path":690,"priority":397},"references/TEXT_GENERATION.md",{"sources":692},[693],"manual",{"closedIssues90d":202,"description":695,"forks":203,"homepage":696,"license":208,"openIssues90d":204,"pushedAt":205,"readmeSize":200,"stars":206,"topics":697},"Give your agents the power of the Hugging Face ecosystem","https://huggingface.co",[],{"classifiedAt":699,"discoverAt":700,"extractAt":701,"githubAt":701,"updatedAt":699},1778690772996,1778689536128,1778690770714,[181,185,180,183,184,182,186],{"evaluatedAt":212,"extractAt":250,"updatedAt":212},[],[706,738,756,789,817,848],{"_creationTime":707,"_id":708,"community":709,"display":710,"identity":716,"providers":721,"relations":731,"tags":734,"workflow":735},1778694269038.6682,"k1752cypc448mke749yjbkc65186mg6f",{"reviewCount":8},{"description":711,"installMethods":712,"name":714,"sourceUrl":715},"This skill should be used when the user asks to \"compress context\", \"summarize conversation history\", \"implement compaction\", \"reduce token usage\", or mentions context compression, structured summarization, tokens-per-task optimization, or long-running agent sessions exceeding context limits.",{"claudeCode":713},"muratcankoylan/Agent-Skills-for-Context-Engineering","Context Compression","https://github.com/muratcankoylan/Agent-Skills-for-Context-Engineering",{"basePath":717,"githubOwner":718,"githubRepo":719,"locale":18,"slug":720,"type":216},"skills/context-compression","muratcankoylan","Agent-Skills-for-Context-Engineering","context-compression",{"evaluate":722,"extract":729},{"promptVersionExtension":173,"promptVersionScoring":174,"score":723,"tags":724,"targetMarket":187,"tier":188},100,[725,182,726,727,728,181],"context-engineering","agent","summarization","compression",{"commitSha":241,"license":730},"MIT",{"parentExtensionId":732,"repoId":733},"k1754dy3wbsv2a5gr1a983zzs586njca","kd7f12maf5nxmx5xttjx7scfnx86m1tv",[726,728,725,181,182,727],{"evaluatedAt":736,"extractAt":737,"updatedAt":736},1778694410149,1778694269038,{"_creationTime":739,"_id":740,"community":741,"display":742,"identity":744,"providers":745,"relations":752,"tags":753,"workflow":754},1778690773482.4866,"k17a3mmgvm5hj49twj487hp64186n2qa",{"reviewCount":8},{"description":439,"installMethods":743,"name":326,"sourceUrl":14},{"claudeCode":12},{"basePath":324,"githubOwner":180,"githubRepo":215,"locale":18,"slug":326,"type":216},{"evaluate":746,"extract":751},{"promptVersionExtension":173,"promptVersionScoring":174,"score":723,"tags":747,"targetMarket":187,"tier":188},[238,180,748,749,750],"mlops","data-management","model-management",{"commitSha":241},{"parentExtensionId":219,"repoId":246},[238,749,180,748,750],{"evaluatedAt":755,"extractAt":250,"updatedAt":755},1778691223210,{"_creationTime":757,"_id":758,"community":759,"display":760,"identity":766,"providers":771,"relations":782,"tags":785,"workflow":786},1778696691708.3308,"k17d3c35ws96bb55ry97apwm5n86mqp2",{"reviewCount":8},{"description":761,"installMethods":762,"name":764,"sourceUrl":765},"Format prompts for different LLM providers with chat templates and HNSW-powered context retrieval",{"claudeCode":763},"ruvnet/ruflo","Chat Format","https://github.com/ruvnet/ruflo",{"basePath":767,"githubOwner":768,"githubRepo":769,"locale":18,"slug":770,"type":216},"plugins/ruflo-ruvllm/skills/chat-format","ruvnet","ruflo","chat-format",{"evaluate":772,"extract":781},{"promptVersionExtension":173,"promptVersionScoring":174,"score":723,"tags":773,"targetMarket":187,"tier":188},[182,774,775,776,777,778,779,780],"prompting","rag","context-retrieval","openai","anthropic","gemini","ollama",{"commitSha":241,"license":730},{"parentExtensionId":783,"repoId":784},"k17ekc0sj70ms9kgkkgr2ypr4s86mz40","kd7ed28gj8n0y3msk5dzrp05zs86nqtc",[778,776,779,182,780,777,774,775],{"evaluatedAt":787,"extractAt":788,"updatedAt":787},1778701390930,1778696691708,{"_creationTime":790,"_id":791,"community":792,"display":793,"identity":799,"providers":803,"relations":810,"tags":813,"workflow":814},1778699234184.611,"k179b6dkc777g1rgyecze04wqn86m6y4",{"reviewCount":8},{"description":794,"installMethods":795,"name":797,"sourceUrl":798},"Process-first advisor routing for Claude, Codex, or Gemini via `omc ask`, with artifact capture and no raw CLI assembly",{"claudeCode":796},"Yeachan-Heo/oh-my-claudecode","oh-my-claudecode","https://github.com/Yeachan-Heo/oh-my-claudecode",{"basePath":800,"githubOwner":801,"githubRepo":797,"locale":18,"slug":802,"type":216},"skills/ask","Yeachan-Heo","ask",{"evaluate":804,"extract":809},{"promptVersionExtension":173,"promptVersionScoring":174,"score":723,"tags":805,"targetMarket":187,"tier":188},[238,806,182,774,807,808],"automation","code-review","artifact-generation",{"commitSha":241,"license":730},{"parentExtensionId":811,"repoId":812},"k17brg5egdw1jbncj1j4wfv3fh86n639","kd74zv63fryf9prygtq7gf4es986n22y",[808,806,238,807,182,774],{"evaluatedAt":815,"extractAt":816,"updatedAt":815},1778699303045,1778699234184,{"_creationTime":818,"_id":819,"community":820,"display":821,"identity":827,"providers":832,"relations":841,"tags":844,"workflow":845},1778696595410.5698,"k171sdysmt658g1cdd7hgt8p8h86nms7",{"reviewCount":8},{"description":822,"installMethods":823,"name":825,"sourceUrl":826},"End-of-session ritual that audits changes, runs quality checks, captures learnings, and produces a session summary. Use when saying \"wrap up\", \"done for the day\", \"finish coding\", or ending a coding session.",{"claudeCode":824},"rohitg00/pro-workflow","Wrap-Up Ritual","https://github.com/rohitg00/pro-workflow",{"basePath":828,"githubOwner":829,"githubRepo":830,"locale":18,"slug":831,"type":216},"skills/wrap-up","rohitg00","pro-workflow","wrap-up",{"evaluate":833,"extract":840},{"promptVersionExtension":173,"promptVersionScoring":174,"score":723,"tags":834,"targetMarket":187,"tier":188},[835,182,836,837,838,839],"workflow","productivity","memory","knowledge-base","code-quality",{"commitSha":241,"license":730},{"parentExtensionId":842,"repoId":843},"k17fxtjcfh5gvxdrhv2dmgn1t986mdhv","kd7am4e918eq98hrd9s31jm4vs86nn0b",[839,838,182,837,836,835],{"evaluatedAt":846,"extractAt":847,"updatedAt":846},1778697164619,1778696595410,{"_creationTime":849,"_id":850,"community":851,"display":852,"identity":856,"providers":859,"relations":867,"tags":868,"workflow":869},1778694269038.6707,"k178ghjhvwyw1pv6vxnaqcwgyx86m2g7",{"reviewCount":8},{"description":853,"installMethods":854,"name":855,"sourceUrl":715},"This skill should be used when the user asks to \"start an LLM project\", \"design batch pipeline\", \"evaluate task-model fit\", \"structure agent project\", or mentions pipeline architecture, agent-assisted development, cost estimation, or choosing between LLM and traditional approaches.",{"claudeCode":713},"Project Development",{"basePath":857,"githubOwner":718,"githubRepo":719,"locale":18,"slug":858,"type":216},"skills/project-development","project-development",{"evaluate":860,"extract":866},{"promptVersionExtension":173,"promptVersionScoring":174,"score":723,"tags":861,"targetMarket":187,"tier":188},[182,862,863,864,865],"project-management","pipeline-architecture","agent-development","batch-processing",{"commitSha":241,"license":730},{"parentExtensionId":732,"repoId":733},[864,865,182,863,862],{"evaluatedAt":870,"extractAt":737,"updatedAt":870},1778694576171]