devantler-tech/agent-skills

A curated index of generic agent skills installable with either the gh skill CLI (v2.90.0+) or npx skills.

These skills are agent-neutral: every SKILL.md follows the agentskills.io spec, so the same skill works in GitHub Copilot, Claude Code, Cursor, Codex, Gemini CLI, and the other agents these CLIs support — pick the target with --agent. See Installing to install for one agent or several at once.

This repo is a pointer list and publisher of in-house skills. Each row below is either an in-house skill or installs directly from its original upstream so gh skill records the true source in the skill's SKILL.md frontmatter (metadata.github-repo, github-path, github-ref, github-tree-sha) and gh skill update --all works natively — no lockfile, no sync bot, no custom metadata.

Skills

GitOps & Kubernetes

Skill	Upstream	Install
gitops-cluster-debug	fluxcd/agent-skills	gh skill install fluxcd/agent-skills gitops-cluster-debug
gitops-knowledge	fluxcd/agent-skills	gh skill install fluxcd/agent-skills gitops-knowledge
gitops-repo-audit	fluxcd/agent-skills	gh skill install fluxcd/agent-skills gitops-repo-audit
gitops-tenant-onboarding	devantler-tech/agent-skills	gh skill install devantler-tech/agent-skills gitops-tenant-onboarding
siderolabs	siderolabs/docs	gh skill install siderolabs/docs siderolabs

GitHub

Skill	Upstream	Install
gh-stack	github/gh-stack	gh skill install github/gh-stack gh-stack
github-actions-docs	xixu-me/skills	gh skill install xixu-me/skills github-actions-docs
github-issues	github/awesome-copilot	gh skill install github/awesome-copilot github-issues

Copilot

Skill	Upstream	Install
copilot-instructions-blueprint-generator	github/awesome-copilot	gh skill install github/awesome-copilot copilot-instructions-blueprint-generator
copilot-sdk	github/awesome-copilot	gh skill install github/awesome-copilot copilot-sdk
find-skills	vercel-labs/skills	gh skill install vercel-labs/skills find-skills

Go

Skill	Upstream	Install
bubbletea	ggprompts/tfe	gh skill install ggprompts/tfe bubbletea --allow-hidden-dirs
golang-pro	Jeffallan/claude-skills	gh skill install Jeffallan/claude-skills golang-pro

Git

Skill	Upstream	Install
git-commit	github/awesome-copilot	gh skill install github/awesome-copilot git-commit

Automated AI Engineer

Skill	Upstream	Install
agent-improvement	devantler-tech/agent-skills	gh skill install devantler-tech/agent-skills agent-improvement
portfolio-maintenance	devantler-tech/agent-skills	gh skill install devantler-tech/agent-skills portfolio-maintenance
product-engineering	devantler-tech/agent-skills	gh skill install devantler-tech/agent-skills product-engineering
self-improvement	devantler-tech/agent-skills	gh skill install devantler-tech/agent-skills self-improvement

Engineering Practices

Skill	Upstream	Install
agent-instructions	devantler-tech/agent-skills	gh skill install devantler-tech/agent-skills agent-instructions
conventional-release	devantler-tech/agent-skills	gh skill install devantler-tech/agent-skills conventional-release
refactor	github/awesome-copilot	gh skill install github/awesome-copilot refactor
test-driven-development	obra/superpowers	gh skill install obra/superpowers test-driven-development
ways-of-working	devantler-tech/agent-skills	gh skill install devantler-tech/agent-skills ways-of-working

Vibe Coding

Skill	Upstream	Install
allowed-stack-guardrail	devantler-tech/agent-skills	gh skill install devantler-tech/agent-skills allowed-stack-guardrail
jargon-free-voice	devantler-tech/agent-skills	gh skill install devantler-tech/agent-skills jargon-free-voice
needs-stack-mapping	devantler-tech/agent-skills	gh skill install devantler-tech/agent-skills needs-stack-mapping

Frontend & Design

Skill	Upstream	Install
astro	astrolicious/agent-skills	gh skill install astrolicious/agent-skills astro
frontend-design	anthropics/skills	gh skill install anthropics/skills frontend-design
web-design-guidelines	vercel-labs/agent-skills	gh skill install vercel-labs/agent-skills web-design-guidelines

Installing

Two CLIs install these skills, and they reach different things — because this repo is a pointer list, not a re-host. Of the 29 rows above, only 11 are in-house (hosted here); the other 18 point at their own upstream repo.

CLI	Reaches	Best for
gh skill	Every row above — each row's command already targets that skill's true source	Anything in the index. Records upstream provenance in the installed SKILL.md, so gh skill update --all works natively. Copy the command from the table, don't retype it against this repo.
npx skills	The 11 in-house skills only — it installs what physically lives in the repo you point it at	The in-house skills, several at once, or an agent gh skill doesn't cover (70+ supported). For an indexed upstream skill, point it at that upstream instead (e.g. npx skills add fluxcd/agent-skills).

Note

There is no registry to sign up for and no package to publish — both CLIs resolve owner/repo straight from GitHub.

With npx skills

npx skills needs no install of its own and prompts for which skills and which agents you want. Pointed at this repo it offers the 11 in-house skills; for an indexed upstream skill, point it at that skill's own repo instead.

Note

Requires Node.js ≥ 22.20.0 (the skills package's declared engines.node). On an older Node this fails before any skill is fetched. gh skill has no Node dependency.

# Browse what's on offer without installing anything
npx skills add devantler-tech/agent-skills --list

# Install specific skills for specific agents
npx skills add devantler-tech/agent-skills --skill ways-of-working --agent claude-code

# Install every in-house skill, for EVERY supported agent, no prompts.
# Note --all is not scoped to agents you have installed — pass --agent to limit it.
npx skills add devantler-tech/agent-skills --all

Add -g to install to your user directory instead of the current project.

The skills.sh directory has no submission step — it lists a repo off anonymous install telemetry from this CLI. That telemetry is opt-out (DISABLE_TELEMETRY or DO_NOT_TRACK) and is disabled automatically in CI, so only telemetry-enabled installs contribute to a listing.

With gh skill

Each gh skill install accepts --agent <name>, --scope user|project, and --pin <ref> (or an @ref suffix on the skill name) — see gh skill install --help for the full list of supported agents.

The install commands in the tables above use the default agent (GitHub Copilot) at project scope. To install for Claude Code instead, or for both agents at once at user scope (so the skill is available everywhere), add --agent / --scope:

# GitHub Copilot, user scope -> ~/.copilot/skills/<skill>/
gh skill install devantler-tech/agent-skills ways-of-working --agent github-copilot --scope user

# Claude Code, user scope -> ~/.claude/skills/<skill>/
gh skill install devantler-tech/agent-skills ways-of-working --agent claude-code --scope user

Install everything for both Copilot and Claude

scripts/install.sh installs every skill listed above for the agents you name (default: github-copilot and claude-code) at user scope:

./scripts/install.sh                          # both Copilot + Claude Code (user scope)
./scripts/install.sh claude-code              # just Claude Code
AGENTS="github-copilot claude-code cursor" ./scripts/install.sh   # any gh skill agents

The script is the single source of truth's consumer — it reads the install commands straight out of this README, so it never drifts from the index.

Automated installation and updates

To adopt these skills in another repository:

• devantler-tech/actions/setup-agent-skills — composite action that installs a newline list of <owner/repo> <skill>[@pin] entries, for one or more agents.
• devantler-tech/actions/update-agent-skills — composite action that runs gh skill update --all against the checked-in skills.
• devantler-tech/actions/.github/workflows/update-agent-skills.yaml — reusable workflow that opens a PR when any skill's upstream has drifted.

All three rely on the github-* metadata that gh skill install injects into each SKILL.md, so no lockfile or external manifest is required.

Contributing

This repository follows the agentskills.io spec: skill directories live at the repository root and include a conformant SKILL.md at their root.

Every pull request runs the 🧪 CI workflow, whose CI - Required Checks aggregator gates the merge on four jobs:

• Publish dry-run — gh skill publish --dry-run confirms the repo is publishable as a skill bundle.
• Spec validation — each in-house skill is validated against the agentskills.io spec with skills-ref validate.
• Script lint — shellcheck over scripts/*.sh.
• Index lockstep — ./scripts/check-readme-index.sh asserts the README ## Skills tables stay in lockstep with every consumer: a non-empty parse, parsed install-count equal to the number of table rows, every in-house skill present in the index, every in-house entry resolving to an on-disk SKILL.md, and each row's Skill/Upstream/Install columns agreeing (so a typo'd repo or slug can't ship a broken install command).

A separate 🔗 Upstream skill targets workflow verifies every upstream row still resolves to a real skill at its source. It runs weekly and on any PR that touches the index, but is deliberately not a required check — a third-party outage must never block a contributor PR, so transient errors downgrade to warnings and only definitive drift fails.

Releases are cut automatically on every push to main — release.yaml uses mathieudutour/github-tag-action to derive the next version tag from commit conventions, and cd.yaml then runs gh skill publish against the resulting tag. The publish pipeline publishes in-house skills (e.g. ways-of-working) on each release.

Inclusion criteria

Before adding a row, check the skill clears the bar this index is curated to. These criteria are already applied in review — they are written here so every contributor (and the autonomous assistant) applies the same bar:

• Generic & reusable. Index a skill only if it is useful beyond a single project or person — a general capability (a framework, a workflow, a language toolchain, an engineering practice). Project- or repo-specific knowledge belongs in that project, not here.
• Upstream pointer by default. Prefer a row that installs directly from the skill's canonical upstream (gh skill install <owner/repo> <skill>) so gh skill update --all tracks the true source — no fork, no copy. Add an in-house directory only for skills devantler-tech genuinely authors and maintains (e.g. ways-of-working).
• Spec-conformant & agent-neutral. The skill's SKILL.md must follow the agentskills.io spec and stay tool-neutral — no Copilot/Claude-only assumptions — so it works across every agent gh skill supports.
• Quality upstream. Point only at an actively-maintained, good-quality source with a precise description (the field agents match on to trigger the skill). Avoid abandoned or low-signal upstreams.
• Naming & category. The row's skill slug matches the upstream skill name; place it under the best-fitting ## Skills category, adding a new category only when a skill clearly fits none of the existing ones.
• Lockstep. Every change updates the README ## Skills tables — the single source of truth that install.sh and the setup-/update-agent-skills consumers parse. Never hand-maintain a parallel list; run ./scripts/check-readme-index.sh — the same gate CI enforces — before opening a PR.

See the devantler-tech organization guidelines for PR/issue templates and general contribution rules.

License

Apache 2.0 — see LICENSE.