alchaincyf/huashu-skills
花叔全部开源 Agent Skills 总目录:16 旗舰 + 14 人物视角 + 22 内置共 52 个 skill,分层分类 + AI Agent 安装协议 + 机器可读 skills.json + 更新检查机制
huashu‑skills – A curated catalogue of Claude‑compatible Agent Skills
What it is
- A single repository that lists 57 “skills” (plug‑in modules) created by the developer Huashu.
- Each skill follows the standard
SKILL.mdformat used by Claude Code, Anthropic’s Codex, Kimi Code and similar AI agents. - The repo itself contains 22 lightweight skills and a machine‑readable index (
skills.json). The remaining skills live in 21 “flagship” repos and 14 persona‑based repos, all linked from this catalogue.
Why it matters
- Modern LLM‑powered agents can be extended with skills – reusable, self‑describing pieces of code that let the agent perform concrete tasks (e.g., generate a design mock‑up, clean Excel data, turn a Markdown file into a PDF, control a native macOS app, etc.).
- Huashu‑skills provides a ready‑made library of such extensions, covering design, data analysis, writing, video creation, code‑speed‑ups, and even “persona” reasoning modules that emulate the thinking style of famous people.
- The catalogue is agent‑agnostic (Claude, Codex, Kimi) and includes an auto‑updater (
huashu-skill-updater) that can check whether installed skills are behind the latest Git tags.
How to use it
- Pick a skill – browse the README tables or the
skills.jsonfile. For example, to generate a PPT from a single sentence you could usehuashu-slides. - Install –
- Standalone skill:
git clone https://github.com/alchaincyf/<repo>.git ~/.claude/skills/<repo> - Built‑in skill: clone this repo once, then copy the desired sub‑directory into
~/.claude/skills/(or a project‑local.claude/skills/).
- Standalone skill:
- Restart the agent – the new skill becomes available in the next session.
- Update check – run
python3 ~/.claude/skills/huashu-skill-updater/scripts/check_updates.pyto see if any installed skill has a newer version.
What you get (selected examples)
| Category | Skill (repo) | What it does |
|---|---|---|
| Design & Visuals | huashu-design (standalone) |
Full‑featured HTML design system for high‑fidelity prototypes, slides, animations, with 20 design philosophies and MP4 export. |
huashu-gpt-image |
Prompt engineering for GPT‑image models – reference‑image‑based prompts, batch grid generation, failure‑mode guide. | |
| Data & Docs | huashu-excel |
End‑to‑end Excel workflow: cleaning, metric calculation, reconciliation, report generation (uses only openpyxl). |
huashu-md-html |
Convert PDFs, DOCX, EPUB, webpages → clean Markdown → publish as polished HTML/DOCX with anti‑AI‑slop themes. | |
| Video | seedance-skill |
AI video workstation driven by “white‑model” references; scene JSON ties prompts to reference footage. |
| Persona Skills | elon-musk-skill, karpathy-skill, … |
Encodes a person’s mental models, decision heuristics and expression style; can be invoked to answer questions “as Elon Musk would”. |
| Utility | huashu-skill-updater |
Scans installed Huashu skills, compares local commit hashes to remote tags, and reports needed updates. |
| Built‑in workflow examples | huashu-slides, huashu-data-pro, huashu-topic-gen |
End‑to‑end pipelines for slide creation, data‑driven reports, and topic generation respectively. |
Who it’s for
- Developers or power‑users who run Claude Code (or compatible agents) and want to extend the agent’s capabilities without writing code.
- Content creators looking for plug‑and‑play pipelines (design, video scripts, article polishing).
- Anyone who wants a single source of truth for installing, updating, and discovering Huashu’s growing skill ecosystem.
Quick start snippet
# Install the design skill (stand‑alone repo)
git clone https://github.com/alchaincyf/huashu-design.git ~/.claude/skills/huashu-design
# Or install a built‑in skill from this repo
git clone --depth 1 https://github.com/alchaincyf/huashu-skills.git /tmp/huashu-skills
cp -r /tmp/huashu-skills/huashu-slides ~/.claude/skills/
# Restart Claude (or your agent) and you can now call the skill
Bottom line – huashu‑skills is a genuine, actively maintained collection of AI‑agent extensions that lets you add concrete, reusable functionality to LLM agents with a single git clone or folder copy. It’s a practical toolbox rather than a mere reading list.
Related
- Project
- Project
- Project
- Project
- Project