スキルを探す

Use when the user needs to plan, de-risk, or ground a software engineering / computer science undergraduate thesis from a real codebase before final writing. Trigger for requests such as "根据项目写毕业论文", "先做论文大纲", "用真实项目材料规划论文", "检查论文是否脱离代码", "根据检测报告降查重", "根据AIGC报告改写定稿", "继续在手改初稿上改", or when an existing draft thesis must be reworked against a school format sample. Produces an evidence-backed outline, chapter rewrite plan, figure plan, and handoff package for academic-paper-composer.

日本語の概要は準備中です。原文の説明を表示しています。

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.

日本語の概要は準備中です。原文の説明を表示しています。

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.

日本語の概要は準備中です。原文の説明を表示しています。

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.

日本語の概要は準備中です。原文の説明を表示しています。

Agents should invoke this skill for academic or technical papers, arXiv/PubMed/IEEE/ACM links, PDFs, methodology review, limitations, practical implications, or extracting findings for engineering decisions.

日本語の概要は準備中です。原文の説明を表示しています。

One effect, done well: 먹 번짐 — a large Korean, Japanese or Chinese display headline that reads as sumi ink printed on hanji. The heading stays real, selectable DOM text in ink color from the first frame; a WebGL2 layer over it adds what real ink does: a wet, glossy stroke with a soft bloom that shrinks as it dries, ink pooling darker at stroke rims, ends and corners, paper fibers showing through thick strokes, and capillary feathering along fibers at a few edges only. Two parameters (bleed, duration), ~5 KiB gzip of library JS, procedural paper texture, static fallback, prefers-reduced-motion, real-GPU verification script. Use for a CJK hero/launch headline, brand title, campaign line or section opener that should feel like a printed calligraphy poster. Triggers: '먹 번짐', '잉크 번짐', '한지 느낌 제목', '한글 타이포 셰이더', '제목에 셰이더', '墨のにじみ', '和紙', '筆文字っぽく', '墨跡暈開', '宣紙', 'CJK headline shader', 'sumi ink text effect', 'ink bleed text', 'Hangul shader'. Not for body text, buttons, or Latin-only headlines. Full-page shader backgrounds → omd-shader-hero. Korean typesetting rules → hangul. Brand tokens → omd-apply.

One effect, done well: 먹 번짐 — a large Korean, Japanese or Chinese display headline that reads as sumi ink printed on hanji. The heading stays real, selectable DOM text in ink color from the first frame; a WebGL2 layer over it adds what real ink does: a wet, glossy stroke with a soft bloom that shrinks as it dries, ink pooling darker at stroke rims, ends and corners, paper fibers showing through thick strokes, and capillary feathering along fibers at a few edges only. Two parameters (bleed, duration), ~5 KiB gzip of library JS, procedural paper texture, static fallback, prefers-reduced-motion, real-GPU verification script. Use for a CJK hero/launch headline, brand title, campaign line or section opener that should feel like a printed calligraphy poster. Triggers: '먹 번짐', '잉크 번짐', '한지 느낌 제목', '한글 타이포 셰이더', '제목에 셰이더', '墨のにじみ', '和紙', '筆文字っぽく', '墨跡暈開', '宣紙', 'CJK headline shader', 'sumi ink text effect', 'ink bleed text', 'Hangul shader'. Not for body text, buttons, or Latin-only headlines. Full-page shader backgrounds → omd-shader-hero. Korean typesetting rules → hangul. Brand tokens → omd-apply.

.omd/preferences.md의 status:pending 항목을 Core v2 System Graph에 합쳐 DESIGN.md를 재생성하고 status를 applied로 플립. '프리퍼런스 정리해줘', 'fold preferences', 'apply all corrections', 「好みをDESIGN.mdに反映」, 「套用偏好」류의 요청에 트리거. 단발성 교정 기록은 omd:remember.

Pre-search topic resolution. Maps vague queries to concrete entities (GitHub orgs, X handles, subreddits, docs URLs) before searching. Run as first step of any research workflow to dramatically improve search precision.

日本語の概要は準備中です。原文の説明を表示しています。

Show how vibecosystem works - agents, skills, hooks, and self-learning pipeline.

日本語の概要は準備中です。原文の説明を表示しています。

Document codebase as-is with thoughts directory for historical context

日本語の概要は準備中です。原文の説明を表示しています。

Research agent for external documentation, best practices, and library APIs via MCP tools

日本語の概要は準備中です。原文の説明を表示しています。

External research workflow for docs, web, APIs - NOT codebase exploration

日本語の概要は準備中です。原文の説明を表示しています。

Query the memory system for relevant learnings from past sessions using semantic search.

日本語の概要は準備中です。原文の説明を表示しています。

Store a learning, pattern, or decision in the memory system for future recall.

日本語の概要は準備中です。原文の説明を表示しています。

Analyze repository structure, patterns, conventions, and documentation for understanding a new codebase

日本語の概要は準備中です。原文の説明を表示しています。

Migration workflow - research → analyze → plan → implement → review

日本語の概要は準備中です。原文の説明を表示しています。

AI-powered web search, research, and reasoning via Perplexity

日本語の概要は準備中です。原文の説明を表示しています。

Hizlica kural/ogrenim kaydet. CLAUDE.md'ye ve memory'ye yazar. Kullanim: /learn <kural>

日本語の概要は準備中です。原文の説明を表示しています。

Interactive workspace discovery - learn what tools, workflows, agents, and hooks are available

日本語の概要は準備中です。原文の説明を表示しています。

Transform session learnings into permanent capabilities (skills, rules, agents). Use when asked to "improve setup", "learn from sessions", "compound learnings", or "what patterns should become skills".

日本語の概要は準備中です。原文の説明を表示しています。

Automatically extract reusable patterns from Claude Code sessions and save them as learned skills for future use.

日本語の概要は準備中です。原文の説明を表示しています。

Instinct-based learning system that observes sessions via hooks, creates atomic instincts with confidence scoring, and evolves them into skills/commands/agents.

日本語の概要は準備中です。原文の説明を表示しています。

Implements onboarding and help systems including product tours, interactive tutorials, tooltips, checklists, help panels, and progressive disclosure patterns. Use when building first-time experiences, feature discovery, guided walkthroughs, contextual help, setup flows, or user activation features. Provides timing strategies, accessibility patterns (keyboard, screen readers, reduced motion), and metrics for measuring onboarding success.

日本語の概要は準備中です。原文の説明を表示しています。