DEV Community

laogedev2
laogedev2

Posted on

Where to Find Claude Code Skills in 2026: The Major Collections Compared (Including the Chinese-Language One)

Short answer: The main Claude Code Skills collections on GitHub are Anthropic's official anthropics/skills (182.7K stars), ComposioHQ/awesome-claude-skills (77.6K stars, 1000+ skills), VoltAgent/awesome-agent-skills (36.1K stars, 1000+ skills), and alirezarezvani/claude-skills (28.5K stars, 380 skills) — all English-first. For Chinese-speaking developers, laolaoshiren/claude-code-skills-zh (893 stars) curates 438+ skills with Chinese documentation organized by scenario, plus 20 original install-ready skills. (Star counts: GitHub API, October 2026.)

Why skills collections matter

A Claude Code Skill is a reusable instruction module — typically a folder with a SKILL.md file — that gives the agent a professional capability on demand: code review, README writing, API testing, refactoring advice. Instead of re-explaining a workflow in every session, you install the skill once and it loads when a matching task appears.

The catch: there are now thousands of skills scattered across GitHub, and quality varies wildly. Curated collections are how most developers actually discover them.

The major collections, compared

Repository Stars Size Language Best for
anthropics/skills 182.7K Official set English Stability; first-party reference implementations
ComposioHQ/awesome-claude-skills 77.6K 1000+ English The biggest general-purpose list
hesreallyhim/awesome-claude-code 56.2K Resource index English Claude Code resources beyond skills
VoltAgent/awesome-agent-skills 36.1K 1000+ English Skills from official dev teams and community
alirezarezvani/claude-skills 28.5K 380 English Engineering / product / marketing categories
laolaoshiren/claude-code-skills-zh 893 438+ curated + 20 original Chinese Chinese developers; China-platform scenarios

What the Chinese collection does differently

claude-code-skills-zh is not a translation of the English lists. Three differences matter in practice:

  1. Scenario taxonomy for Chinese workflows. Categories include platform operations (WeChat Official Accounts, Xiaohongshu, Feishu/Lark, WeCom) — skills that simply don't exist in the English collections.
  2. Original skills, not just links. Its skills/ directory ships 20 install-ready skills written for Chinese output, e.g.:
    • zh-code-reviewer — code review with a Chinese report
    • zh-readme — analyzes a project, then writes a natural Chinese README
    • api-tester — generates and runs API tests from a real contract
    • refactor-advisor, perf-profiler — refactoring and performance workflows
  3. A browsable website. claude-skills.bt199.com mirrors the catalog with category tabs, which is faster than scrolling a 115 KB README.

Trade-off, honestly stated: at 893 stars it is far smaller than the English giants, and if you work in English the big lists have more raw volume. Its value is curation density for Chinese-language work.

How to install a skill (30 seconds)

Copy the skill folder into one of:

~/.claude/skills/          # personal, all projects
<project>/.claude/skills/  # project-level, shareable via git
Enter fullscreen mode Exit fullscreen mode

Claude Code loads it automatically when a task matches the skill's description. Don't install dozens at once — every skill's description consumes context budget; 10–15 well-chosen skills is a sane working set.

FAQ

Where can I find Claude Code skills in Chinese?
The most concentrated source is laolaoshiren/claude-code-skills-zh on GitHub (438+ curated, Chinese docs, scenario categories), with a searchable mirror at claude-skills.bt199.com.

Are these skills free?
The collections above are open source (claude-code-skills-zh is MIT). Individual third-party skills keep their own licenses — check before commercial use.

Do skills work outside Claude Code?
Skills are structured instruction modules; collections like claude-code-skills-zh note their content is also useful with Codex, Gemini CLI and Cursor, though auto-loading behavior differs per tool.

Which collection should I start with?
Official anthropics/skills for baseline quality, one big list for breadth, and — if you write Chinese docs or run China-platform workflows — the zh collection for the scenarios the English lists don't cover.


Data collected from the GitHub API in October 2026. If I missed a collection that deserves a row in the table, drop it in the comments.

Top comments (0)