Generate Hugging Face Hub (huggingface_hub) release notes from cached PR JSON files. Use when asked to draft release notes from PR files.
npx skills add https://github.com/huggingface/huggingface_hub --skill hf-release-notes
Generate release notes for huggingface_hub from cached PR JSON files. This skill reads PR metadata, categorizes entries, and produces a formatted markdown release notes document.
Output directory: The prompt will specify the output directory as <output_dir>. All paths below use this placeholder.
Read all PR JSON files from <output_dir>/tmp/pr_*.json. Each file contains:
{
"number": 1234,
"title": "...",
"author": "username",
"merged_at": "2026-01-15T10:30:00Z",
"body": "...",
"labels": ["highlight", "cli"],
"url": "https://github.com/huggingface/huggingface_hub/pull/1234",
"doc_diffs": [
{
"filename": "docs/source/en/guides/cli.md",
"status": "modified",
"patch": "@@ -10,6 +10,10 @@ ..."
}
]
}
The doc_diffs field contains unified diffs for any .md files under docs/ that were
changed in the PR. This is empty ([]) for PRs with no documentation changes.
A PR should be highlighted if:
"highlight" label on GitHub, orfeature, a meaningful UX improvement, or a notable breaking change) even without the label.
Use your judgment — not every feature PR needs a highlight, but don't limit highlights
to only labeled PRs. Each highlight section follows this structure:
## 🖥️ New CLI commandsAvoid bullet points here; write natural sentences. Only use bullets if the content
truly calls for a list (e.g., enumerating 4+ distinct sub-features).
block with a concrete usage example (cherry-pick from doc diffs when available).
- PR title by @author in #1234
For non-highlight PRs, classify into sections using references/sections.md heuristics based on:
For highlighted PRs and other PRs that introduce new features, commands, or APIs:
Use doc diffs first:
doc_diffs field in the PR JSON. If present, these contain the actualdocumentation changes made in the PR (unified diff format).
more accurate summaries. The diffs show exactly what the PR author wrote in the docs.
filename field maps to a docs page URL. For example,docs/source/en/guides/cli.md → https://huggingface.co/docs/huggingface_hub/main/en/guides/cli
Fetch full doc pages when needed:
https://huggingface.co/docs/huggingface_hub/main/en/index
relevant. For example:
https://huggingface.co/docs/huggingface_hub/main/en/guides/clihttps://huggingface.co/docs/huggingface_hub/main/en/guides/inferenceHow to reference:
📚 Documentation: Guide name
- PR title by @author in #1234 — docs
Output to <output_dir>/RELEASE_NOTES_<version>.md using the structure from references/release-notes-template.md:
# [vX.Y.Z] <tagline> (derive tagline from main highlights; always use the base version without prerelease suffix, e.g. [v1.8.0] even when generating for v1.8.0.rc0)Before finishing:
<output_dir>/tmp/ appears exactly once<output_dir>/tmp/ — those PRs belong to a different release and must not appear in these notesby @author in #1234<output_dir>/tmp/<output_dir>/RELEASE_NOTES_<version>.mdreferences/release-notes-template.md: Skeleton structure for release notesreferences/sections.md: Keyword-based section mapping and guidancehttps://huggingface.co/docs/huggingface_hub/main/en/index> Find Earth2Studio models, data sources, and examples for a weather/climate use case. Do NOT use for writing inference code, downloading data, or installation.
> Guide installing Earth2Studio via uv or pip, selecting model extras, and configuring the environment. Do NOT use for writing inference code, choosing models, or PhysicsNeMo questions.
Tokenize, tag, and analyze natural language text using Apple's NaturalLanguage framework and translate between languages with the Translation framework. Use when adding language identification, sentiment analysis, named entity recognition, part-of-speech tagging, text embeddings, or in-app translation to iOS/macOS/visionOS apps.
Routes any legal task to the right LLM, like OpenRouter but for legal work and grounded in benchmarks instead of brand loyalty. Built from mid-2026 legal evals (legalbenchmarks.ai, Vals AI × Stanford LegalBench across 124 models, Harvey's Legal Agent Benchmark, the Atticus Project's CUAD/MAUD/ACORD) plus translation evidence (WMT25, SwiLTra-Bench, ArabLegalEval). Covers five verticals: contract drafting, info extraction, legal research, contract review, and legal translation (including Arabic/MENA). Each asks up to four questions (cost, speed, accuracy/stakes, privacy/jurisdiction/language), then returns a primary model, a fallback, what to avoid, and what a human must verify. Core principle: capability is not controllability, so every route ends with a verification step. Not legal advice; a lawyer owns the output.
Generates standalone interactive HTML "deal cards" that translate complex regulations into negotiation-ready reference tools, systematically distinguishing mandatory obligations from negotiable implementation choices. Use when the user needs an interactive regulatory guide for (1) contract negotiation support, (2) client education or internal training, (3) regulatory briefings for commercial stakeholders, or (4) structured comparison between required and flexible compliance paths. Primary focus on EU digital regulation (Data Act, AI Act, CRA, DORA, NIS2, GDPR) but the structural pattern transfers to any regulation where separating hard obligations from implementation choice is the point. Supports bilingual output where the jurisdiction calls for it.
> Pick the right LLM for CONTRACT DRAFTING — generating, redlining, or rewriting contract language from instructions. Vendor-neutral routing grounded in mid-2026 legal benchmarks (legalbenchmarks.ai Contract Drafting). Asks up to 4 quick questions (cost, speed, accuracy/ stakes, privacy/jurisdiction/language), then recommends a primary model + fallback + what to avoid + what a human must verify. Use when someone asks "which model should I use to draft this clause/agreement", "best AI for drafting contracts", "route this drafting task", or is about to generate/redline contract text and hasn't fixed a model.
Draft matter status reports from emails, call notes, and updates. Internal and client-facing formats, RAG logic, variance commentary, escalation flags. Use when asked to draft a status report, write a project update, summarise matter progress, prepare a client report, create a weekly or monthly update, convert emails into a status summary, or produce any kind of matter reporting. Also triggers when the user pastes email threads and asks what the status is, or needs to turn internal updates into client-facing reports.
Use as stage 2 of the Butterbase journey, after journey-idea has written 01-idea.md. Translates the idea + capability map into a concrete Butterbase plan — tables (with columns/types/RLS shape), auth providers, function list (name + trigger), storage buckets, AI/RAG/realtime/durable usage, and the chosen frontend stack. In hackathon mode, ruthlessly cuts scope into a "ship now" vs "post-hackathon" split. Produces docs/butterbase/02-plan.md.
Take huggingface/hf-release-notes from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.