OpenLink / Open Skills

Open skills for serious agent work.

Discover, inspect and install reusable skills for OpenLink and the coding agents you already use. Every capability starts as readable source.

Explore on GitHub Browse the catalog
npm install --global @zorker/skills

MIT licensed · Node 20+ · zero runtime dependencies

zorker-skills / catalogverified
Search open skills2 results
AI Website Strategistai-website-strategist01
Zorker Demo Remotionzorker-demo-remotion02

$ai-website-strategist audit this product site

Installed skill ready to invoke

Illustrative OpenLink surface using the public Zorker Skills catalog.

Public catalog / v1

Small by design. Complete by default.

Each entry points to its full instructions, references and tools—not a closed prompt hidden behind a product surface.

v1.0.0
AI Website Strategist

$ai-website-strategist

Plan, score, scaffold and validate complete public websites and AI product portals.

StrategySEO / GEOUXValidation
zorker-skills install ai-website-strategist --agent all
skills/ai-website-strategistInspect source
v1.0.0
Zorker Demo Remotion

$zorker-demo-remotion

Direct cinematic product films and tutorials with real source components and deterministic production checks.

RemotionProduct demoCinematographySource UI
zorker-skills install zorker-demo-remotion --agent codex
skills/zorker-demo-remotionInspect source

From source to session

A capability path you can follow.

  1. 01

    Discover

    Browse a small, machine-readable catalog.

  2. 02

    Inspect

    Read the complete SKILL.md before installation.

  3. 03

    Install

    Target one agent or every detected agent.

  4. 04

    Invoke

    Reference the installed skill directly with $.

Inside OpenLink

Skills define the method. OpenLink keeps it in context.

Installed skills are discovered from the active agent runtime and become searchable prompt resources. Reference one with $ without leaving the project session.

Project composer Codex skills
$ai-website-strategistaudit this launch page before release
Skill source and project files remain inspectableEnter ↵

Models remain your choice

Method and intelligence stay separate.

A skill describes how work should be approached. Your configured model provides the intelligence behind that work, so teams can change providers without rewriting the skill.

Session-scoped model selection
Configured provider credentials
No silent fallback to another model
See how agents run

Open by design

Trust starts before installation.

Read the instructions, validate the package and decide exactly where a skill belongs.

Readable source

Every included skill lives in a public repository with its complete entrypoint and references.

Safe installation

Differing installations are preserved unless an explicit force update is requested.

Validated catalog

Catalog integrity, portable files and package contents are checked before publishing.

Build in public

Use a skill. Inspect it. Improve it.

Start with the public catalog or contribute a focused capability that other agent workflows can reuse.

Open the repository Contribute a skill

One project. Every path to production.