shengyy/agent-skills
Versioned, CI-validated agent skills for Claude Code and Codex, installed with one command.
About
Reusable skills for Claude Code and Codex built on the open Agent Skills format, installable with a single npx skills command — either the whole pack or individual skills like codex-construction. The repo is versioned with releases, ships a validate-skills CI workflow, and supports global or per-project installs.
The individual skill contents are less documented than the packaging, so judge by the skill you actually need. Install and update go through the skills CLI, which keeps maintenance simple.
What it can do
- One-command install of the full skill pack or single skills
- Global or per-project install scopes
- Versioned releases
- CI-validated SKILL.md format
- Cross-agent compatibility (Claude Code, Codex, Cursor)
Setup
- Install all: npx skills add shengyy/agent-skills -g --all
- Install one: npx skills add shengyy/agent-skills -g --skill codex-construction
- List available: npx skills add shengyy/agent-skills -l
- Update: npx skills update -g
Prerequisites
Node 18+ with npx; a supported agent harness
Use cases
Works with
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: versioned releases + validate-skills CI is better hygiene than most skill repos; one-command install via npx skills with per-skill selection; individual skill quality varies and is thinly documented; star count not verified — metrics tier. License: MIT (stated in repo badge).
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 6.0/10
- Maintenance · recency, releases, issue load
- 7.0/10
- Openness · license + docs
- 8.0/10
- Manual review · sanity read + red-flag check
- 6.0/10
- · versioned releases + validate-skills CI is better hygiene than most skill repos
- · one-command install via npx skills with per-skill selection
- · individual skill quality varies and is thinly documented
- · star count not verified — metrics tier
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.