getskillpack · skillget

Skill manager & registry for agent and IDE skills

Built to fit the developer workflow you already have: install a vetted skill in one command, pin versions in your repository, and reproduce the same setup for your team and agents — without zip bundles, ad-hoc runbooks, or replacing npm, pip, or Composer for your code.

Same muscle memory as npm · PyPI · Packagist: search → install → lockfile — applied to skills (manifest + tarball) with an open HTTP registry contract.

North star. We are building in the open so teams can trust, fork, and contribute. Rough compass for the org: on the order of 10k cumulative GitHub stars across our flagship public repositories over the next couple of years — a directional signal of adoption, not a quarterly target baked into the product roadmap.

Flagship repos to watch: cli, registry, skillget-manager — starring helps discovery; we never use bots or reciprocal schemes (see playbook below).

How it works

From the registry to a pinned install tree in your repo.

Registry HTTP API and skill metadata
skillget (CLI) search, install, config
skills.lock pinned versions
.skillget/ local install tree

Quick start

Default registry base: https://registry.skpkg.org/api/v1 (override when needed). Full step-by-step (npm or Go, verification, troubleshooting): Zero to first skill.

One command try (Node 18+):

# Fast path — global CLI, then hit the public registry
npm install -g @getskillpack/cli && skillget search

From source (Go) — build in ~five minutes:

# Build the Go client
git clone https://github.com/getskillpack/cli.git && cd cli
go build -o skillget ./cmd/skillget

# Local or custom registry
export SKILLGET_REGISTRY_URL=http://localhost:3000/api/v1

./skillget search
./skillget install <skill-name>

After install, the CLI updates skills.lock and writes artifacts under .skillget/skills/<name>/<version>/.

Signals we track (funnel + trust)

We review these on a weekly cadence for the org (dashboard or short doc for board) — not on every automation heartbeat. Traffic and clone charts come from GitHub Insights per repository where enabled.

Signal What it tells us Primary source
GitHub stars Discovery and trust in the open; rolls up toward the long-horizon org compass above. Repo stargazer counts (e.g. cli, registry)
Clones & visitors Interest before someone stars or installs; highlights README and landing effectiveness. GitHub → Insights → Traffic (per repo)
Referrers / site traffic Which surfaces send people to the README, releases, or this landing (when Pages analytics are available). GitHub Traffic + Pages (as configured)
Releases & tags Predictability for early adopters; pairs with changelog quality. Releases tab per repository

Full playbook, anti-spam rules, and handoff checklist: MARKETING_LANDING_AND_GROWTH.md (English) · Russian mirror for internal notes.

Trust and transparency

License and source

CLI and related packages in the org ship as open source (see LICENSE in each repo). We do not publish a public SLA before GA — timelines and priorities show up in Issues and releases.

Roadmap

Follow work and releases across the getskillpack organization: regular tags on cli, registry, and skillget-manager are the adoption signal for early users.

For skill authors

Publishing contract and metadata formats live in the registry repository. Start from registry-api.md and the getskillpack/registry sources.