Lizo's Skills Marketplace

   

A skills marketplace authored and maintained by Liz Osborn (Lizo-RoadTown).

The skills follow the open Agent Skills standard (originally from Anthropic), which means they work across any compatible agent — Claude Code, Cursor, Gemini CLI, OpenCode, Junie, Goose, Amp, and others. This repo also ships the Claude Code plugin marketplace manifest so the skills install with one command in Claude Code.

Both skills are referenced by project-starter's variant templates. Until now, they had no public install path — this repo fixes that.

Plugins in this marketplace

liz-patterns ⭐ canonical patterns library

The ONE home for Liz's reusable agents + skills + tools. Installed once, available in every Claude Code session in every project — so the same pattern name means the same thing wherever invoked.

Per tapestry/MANIFESTO.md Part 3 Pillar 1, every reusable pattern has ONE name, ONE home, available everywhere via reference (not copy). This plugin replaces the duplicated copies that previously lived across docs-agent, Make_Skills, and the-loom.

Contains:

• 7 agents — agentic-upskilling (doc wrapper for the deployed self-observer service), eval-deep-research, infrastructure-mapping, lessons-learned, next-actions-planning, orchestration-cataloging, web-app-scaffold
• 8 skills — agentic-skill-design, deep-research-pattern, design-evaluation, document-parsing, documentation, layered-explanation, open-source-documentation, proposal-authoring

Compiled by the platform's recursive-skill-engine loop: candidates surface via observers, decisions land via policy, engine compiles, results land here.

make-skills-discipline

Auto-injecting discipline wrapper for Claude Code sessions working in Make_Skills or any project-starter-scaffolded repo. Hook scripts enforce PROBE-first behavior, file:line citation, dev-tooling-vs-runtime distinction, and friction-as-memory writing. Same forcing-function pattern as superpowers:using-superpowers.

onboarding-psychologist

A behavioral-research framework for designing first-time-user experiences using the IDENTITY-TO-HABIT arc:

1. Define the first win 2. Remove unnecessary setup 3. Create ownership through action 4. Attach a stable cue 5. Reinforce identity

Grounded in BJ Fogg's Tiny Habits and James Clear's identity-based habits. Activates on signup flows, empty states, welcome screens, and reactivation work. Refuses to produce feature tours, "12 things to know" walls, and CRUD-form empty states.

ai-agents-architect

A decision framework for autonomous-agent architecture:

• Where on the autonomy spectrum does this agent sit?
• Which loop pattern — ReAct, Plan-and-Execute, or Tree-of-Thoughts?
• Single agent or multiple? (Most common mistake: splitting agents without a real reason.)
• When to introduce an orchestrator?

Activates on agent-design and orchestration questions. Refuses tool overload, infinite tool loops without a cap, and unverified multi-agent splits.

Install

Claude Code

In an active Claude Code session, run:

/plugin marketplace add Lizo-RoadTown/claude-skills-marketplace
/plugin install liz-patterns@lizo-skills
/plugin install make-skills-discipline@lizo-skills
/plugin install onboarding-psychologist@lizo-skills
/plugin install ai-agents-architect@lizo-skills

Verify with /plugin list.

Most important: liz-patterns is the canonical patterns library. Installing it makes every reusable agent + skill available in every project, with one name, behaving consistently. If you install nothing else, install that.

Other Agent Skills clients (Cursor, Gemini CLI, OpenCode, etc.)

Each client has its own way of registering skills; the Agent Skills client showcase links to each one's docs. The skills + agents are at:

• plugins/liz-patterns/skills/ (8 skills)
• plugins/liz-patterns/agents/ (7 agents)
• plugins/onboarding-psychologist/skills/onboarding-psychologist/
• plugins/ai-agents-architect/skills/ai-agents-architect/
• plugins/make-skills-discipline/ (skills + hooks)

You can clone this repo and point your client at those directories, or copy the individual skill folders into your client's skills directory (usually something like ~/.<client>/skills/).

Validation

Validate the skill format with the official reference tool from agentskills.io:

npx skills-ref validate ./plugins/onboarding-psychologist/skills/onboarding-psychologist
npx skills-ref validate ./plugins/ai-agents-architect/skills/ai-agents-architect

Repo layout

claude-skills-marketplace/
├── .claude-plugin/
│   └── marketplace.json                          # the catalog (4 plugins)
├── plugins/
│   ├── liz-patterns/                             # ⭐ canonical patterns library
│   │   ├── .claude-plugin/
│   │   │   ├── plugin.json
│   │   │   └── marketplace.json                  # dev marketplace for local testing
│   │   ├── agents/                               # 7 agents (description + capabilities + tools)
│   │   ├── skills/                               # 8 skills (name + description)
│   │   └── README.md
│   ├── make-skills-discipline/
│   │   ├── .claude-plugin/plugin.json
│   │   ├── hooks/                                # discipline-enforcement hooks
│   │   ├── skills/
│   │   ├── agents/
│   │   ├── commands/
│   │   └── scripts/
│   ├── onboarding-psychologist/
│   │   ├── .claude-plugin/plugin.json
│   │   └── skills/onboarding-psychologist/SKILL.md
│   └── ai-agents-architect/
│       ├── .claude-plugin/plugin.json
│       └── skills/ai-agents-architect/SKILL.md
├── README.md
└── LICENSE

Follows the standard Claude Code marketplace structure documented at <https://docs.claude.com/en/docs/claude-code/plugin-marketplaces>.

Updating

Bump the version field in plugins/<name>/.claude-plugin/plugin.json and the matching entry in .claude-plugin/marketplace.json when releasing a new version. Push to main. Users get updates when they run /plugin marketplace update followed by /plugin update.

Contributing

Contributions welcome — bug reports, content improvements, new skill proposals. See CONTRIBUTING.md for conventions and the PR flow.

By contributing you agree your contribution is licensed Apache 2.0. Please follow the Code of Conduct.

Security

Report vulnerabilities privately — see SECURITY.md. Don't open public issues for security problems.

Citing this work

If you use these skills in research or downstream projects, see CITATION.cff — GitHub renders it as a "Cite this repository" button.

Author & maintainer

Liz Osborn (Lizo-RoadTown)

• Related: project-starter — the day-1 project scaffolder that references these skills
• Related: Make_Skills — Liz's broader skills workshop

License

Apache 2.0 — see LICENSE.