메뉴
Discover GitHub repositories, find example scripts within them, and read source files with line-range support and automatic Jupyter notebook conversion. Use BEFORE writing any ML training, fine-tuning, or inference code — your knowledge of library APIs is outdated and example scripts show the current patterns.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Alternative paper search backends — OpenAlex (~250M works, no auth, strong PDF extraction) and CORE (~49M full-text, BODY search). Use when hf-papers can't find a paper, when Semantic Scholar is rate-limited, when you need PDFs of older/long-tail papers, or when you need to search the actual body text of papers (not just titles/abstracts).
Inspect a HuggingFace dataset in one call — status, configs/splits, schema, sample rows, parquet file listing, and detailed messages-column analysis (roles, keys, tool-calls). REQUIRED before submitting any training job to verify column format matches the training method.
Explore and fetch HuggingFace library documentation (transformers, trl, datasets, peft, accelerate, trackio, vllm, optimum, courses, etc.). Use when you need current trainer arguments, config parameters, or API reference that your internal knowledge may have wrong.
Submit and manage HuggingFace compute jobs — run training / fine-tuning / inference / batch work on HF cloud infrastructure in Python or Docker mode. Supports one-shot and recurring (cron-style) jobs.
Discover ML research papers, read their contents by section, trace citations, find linked datasets/models — combining HuggingFace Papers, arXiv, and Semantic Scholar. Use whenever you need to research a method, find training recipes with published results, crawl a citation graph, or locate the datasets a paper used. This is your starting point for any literature-first workflow.
File and git-like operations on HuggingFace repositories (model / dataset / space). Covers list / read / upload / delete of files, branches, tags, pull requests, and repo management (create / update). Use to inspect repos, ship artifacts, and propose changes via PRs.
| name | github-code |
| description | Discover GitHub repositories, find example scripts within them, and read source files with line-range support and automatic Jupyter notebook conversion. Use BEFORE writing any ML training, fine-tuning, or inference code — your knowledge of library APIs is outdated and example scripts show the current patterns. |
Three scripts:
scripts/list_repos.py — list and sort a user/org's repositories.scripts/find_examples.py — find example/tutorial files in a specific repo using fuzzy directory matching (examples/, scripts/, notebooks/, cookbook/, …).scripts/read_file.py — read file contents with line ranges; .ipynb → Markdown conversion.MANDATORY before writing ML training / fine-tuning / inference code. Your internal knowledge WILL hallucinate imports, trainer arguments, and config parameters. Working examples show current APIs.
Also useful when:
list_repos on huggingface, pytorch, etc.).--sort updated).# 1. Which libraries does an org have?
python scripts/list_repos.py --owner huggingface --sort stars --limit 10
# 2. What examples exist in the chosen repo?
python scripts/find_examples.py --repo trl --keyword sft
python scripts/find_examples.py --repo transformers --keyword 'flash attention' --max-results 20
# 3. Read the implementation
python scripts/read_file.py --repo huggingface/trl --path examples/scripts/sft.py
python scripts/read_file.py --repo huggingface/transformers \
--path src/transformers/models/llama/modeling_llama.py \
--line-start 300 --line-end 450
list_repos.py:
--owner (required), --owner-type {user|org}--sort {stars|forks|updated|created}, --order {asc|desc}, --limit Nfind_examples.py:
--repo (required), --org (default huggingface)--keyword (optional fuzzy match), --max-results N, --min-score N (0–100)read_file.py:
--repo owner/repo (required), --path (required)--ref (branch / tag / commit, default HEAD)--line-start, --line-end (1-indexed inclusive)All accept --json.
GITHUB_TOKEN — required. A fine-grained token with public-repo read scope is enough.read_file.py detects .ipynb automatically, parses the notebook JSON, strips outputs, removes cells tagged hide/hidden/remove, and converts code+markdown cells to Markdown. Install nbformat + nbconvert (already in pyproject.toml) or the raw JSON is returned untouched.
hf-papers — use find_datasets / find_models on a paper, then jump here to find the implementation.hf-docs — use alongside: github-code gives you a working recipe, hf-docs gives you the authoritative API reference.