ClaudeSkills 1.0.0
The first stable ClaudeSkills release: 13 curated, inspectable Agent Skills built on the open SKILL.md format, with four flagship paths for research, product decisions, decks, and WeChat content production.
ClaudeSkills is a cross-client Agent Skills collection, not a Claude Code-only package. The current installer uses the .agents/skills/ convention by default and keeps Claude Code's native .claude/skills/ path as an explicit compatibility target.
Highlights
- Rebuilt the bilingual GitHub Pages site around immediate installation, workflow inputs and outputs, real repository artifacts, and explicit trust boundaries.
- Added
VERSIONas the canonical repository release version and synchronized both READMEs, site metadata, changelog, and Release links. - Kept the site dependency-free and safe under the
/ClaudeSkills/GitHub Pages project path. - Added release-version and cross-client-positioning validation, including negative tests for drift and legacy Claude Code-only wording.
Verification
- 13 curated skill structures: pass
- 91 routing case definitions across 10 skills: pass
- GitHub Pages validator and 8 unit tests: pass
- Responsive browser checks at 320, 360, 390, 768, 1024, and 1440 px: pass on both languages
These checks do not claim production E2E coverage for every skill, live provider availability, model-executed routing accuracy, or third-party certification. See the site evidence ledger and SECURITY.md for the exact boundaries.