本文へ移動
cccskills
無料GitHub で公開

plugin-marketplace-setup

Designs public or private Agent Skill and plugin marketplaces for Cursor, Claude Code, Codex, Zed, Open Plugin, and npx skills—manifest layout, install matrix, and Skill Steward vs product boundaries. Use when setting up a marketplace, distributing skills/plugins to a team, private registry, .cursor-plugin, .codex-plugin, .plugin, .claude-plugin, or skills.sh publishing.

インストール方法を見る

含まれるファイル(10)

  • SKILL.md13.7 KB
  • evals/cases/adr-plan-dormant.yaml392 B
  • evals/cases/local-copy-pattern-trigger.yaml921 B
  • evals/cases/private-marketplace-trigger.yaml470 B
  • evals/cases/skills-only-install-dormant.yaml459 B
  • references/distribution-matrix.md2.1 KB
  • references/evals.md1.2 KB
  • references/local-copy-pattern.md5.5 KB
  • references/manifest-snippets.md4.2 KB
  • references/sources.md1.6 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Plugin & skill marketplace setup

Ship skills (portable instructions), plugins (runtime wiring), and marketplaces (catalogs) without mixing them up.

Three layers (normative)

LayerWhat it isInstall unitSkill Steward home
SkillSKILL.md per agentskills.ionpx skills add owner/repo --skill nameskills/{name}/
PluginOpen Plugin or vendor manifest + hooks/MCP/rules/commandsPer-agent (see matrix)plugins/{id}/ (ADR 0004)
MarketplaceCatalog of plugins/skillsAdd marketplace, then install pluginProduct repos; Skill Steward uses public Git + skills.sh

Rule: Instructions only → skill. Event hooks or multi-file wiring → plugin (references skills by id, do not fork SKILL.md).

Skill Steward meta-plugin means: a declarative manifest, referenced skill IDs, hook/rule snippets, and validated wiring artifacts. It does not mean a host marketplace submission has happened, and it must not silently mutate Cursor/Codex/Claude configuration. Host-specific install steps remain explicit in README/docs.

Public vs private (decision)

GoalRecommended channelNotes
Public discoveryGitHub public repo + skills.shnpx skills add owner/repo; index needs valid skills/*/SKILL.md
Team-only skills (any agent)Private GitHub/GitLab repoSame npx skills add org/private-repo if agents can clone; use deploy keys / SSO
Cursor team pluginsTeam marketplace — Dashboard → Settings → Plugins → Import repoTeams/Enterprise; private GitHub repo with .cursor-plugin/marketplace.json
Claude Code teamPrivate git marketplace/plugin marketplace add with token env for auto-update (Claude docs)
Codex team$REPO_ROOT/.agents/plugins/marketplace.json or ~/.agents/plugins/marketplace.jsonRepo/personal marketplace; install from app or CLI
Zed skills.agents/skills/ or ~/.agents/skills/Skills only; Zed has no remote skill registry or custom search paths
npm-private skills packagesClaude marketplace source.type: npmFor npm-hosted plugin bundles; uncommon for plain SKILL.md trees

Skill Steward default: public arenukvern/skill_steward for meta-skills; hooks documented separately (npx skills does not install Cursor hooks).

Cross-agent install matrix (skills)

Use vercel-labs/skills as the shared installer:

npx skills add owner/repo                    # all skills, project scope
npx skills add owner/repo --skill my-skill   # one skill
npx skills add owner/repo -a cursor -a claude-code -a codex -a zed -y
npx skills add owner/repo -g                 # global (~/.agents/skills/)
npx skills add owner/repo --copy             # copy instead of symlink

Discovery also reads .claude-plugin/plugin.json and .claude-plugin/marketplace.json skill paths (Claude marketplace compatibility).

Agent--agent flagProject skills pathGlobal skills path
Cursorcursor.agents/skills/ (Cursor also reads .cursor/skills/)~/.cursor/skills/
Claude Codeclaude-code.claude/skills/~/.claude/skills/
Codexcodex.agents/skills/~/.codex/skills/
Zedzed.agents/skills/~/.agents/skills/
Copilot, Windsurf, Cline, …see npx skills --helpper CLI tableper CLI table

Hooks: supported on Claude Code, Cline, Kiro—not Cursor via npx skills. Cursor hooks need .cursor/hooks.json (Skill Steward example).

Cross-agent install matrix (plugins)

AgentManifest dirMarketplace fileInstall (typical)
Cursor.cursor-plugin/plugin.json.cursor-plugin/marketplace.json (multi-plugin repo)Marketplace UI, team import, or init script → .cursor/plugins/local/
Claude Code.claude-plugin/plugin.json.claude-plugin/marketplace.json/plugin marketplace add owner/repo then /plugin install name@marketplace
Codex.codex-plugin/plugin.json$REPO_ROOT/.agents/plugins/marketplace.json or ~/.agents/plugins/marketplace.jsonApp plugin directory or codex plugin add name@marketplace
Open Plugin v1.plugin/plugin.jsonhost-specificPortable baseline; add vendor manifests only for host-specific overrides
Open skills onlyN/AN/Anpx skills add — no MCP/hooks

Reference product layout: product repository marketplace distribution configs.

Workflow: repo-local copy/init pattern

Use this when a product repo needs agents to generate a local installer like flutter-mcp-toolkit init: a repo-owned script that copies canonical skills, host plugin manifests, MCP config, hooks, assets, and marketplace catalog files into Codex/Cursor/Claude layouts.

  1. Inspect source truth — find canonical skills/, existing plugins/, .mcp.json/mcp.json, hooks, package/version metadata, assets, and current install docs. Do not generate host manifests from memory alone.
  2. Choose the channel — use npx skills add --copy for skills-only distribution; use a repo-local copy/init script when skills must ship with MCP, hooks, commands, or assets; use host marketplace manifests only when the repo needs Codex/Cursor/Claude plugin install flows.
  3. Create a product-owned plugin payload — keep canonical skills in the product repo, then copy or generate the install payload under plugin/ or plugins/{id}/ with .codex-plugin/, .cursor-plugin/, .claude-plugin/, optional .plugin/, skills/, hooks, MCP config, and assets.
  4. Write explicit local scripts — generate an idempotent script owned by the product repo (tool/install_agent_bundle.dart, bin/<tool> init, or equivalent). The script should plan/copy directories, create host manifest directories, write marketplace root files, and report changed paths.
  5. Document exact install commands — include separate Codex, Cursor, Claude, and skills-only commands. Name cache refresh/reinstall behavior and rollback steps.
  6. Keep claims bounded — "local copy pattern implemented" does not mean public marketplace submission, host review, runtime MCP correctness, or fresh-agent proof.

See local-copy-pattern.md for a reusable layout, script skeleton, and validation checklist.

Codex plugin notes

  • Keep only plugin.json in .codex-plugin/; put skills/, .mcp.json, .app.json, hooks/, and assets/ at plugin root.
  • Include skills, mcpServers, apps, or hooks in .codex-plugin/plugin.json only when the companion files/directories exist. Codex also checks default hooks/hooks.json.
  • Marketplace files live at $REPO_ROOT/.agents/plugins/marketplace.json for repo/team catalogs or ~/.agents/plugins/marketplace.json for personal catalogs.
  • Each marketplace entry must include policy.installation, policy.authentication, and category; keep source.path relative to the marketplace root and inside that root.
  • Codex installs marketplace plugins into ~/.codex/plugins/cache/$MARKETPLACE_NAME/$PLUGIN_NAME/$VERSION/; local source edits require reinstall/cache refresh before new threads see them.

Open Plugin portability

  • Use .plugin/plugin.json when a plugin should be vendor-neutral across Open Plugin hosts.
  • Add .codex-plugin/plugin.json, .cursor-plugin/plugin.json, or .claude-plugin/plugin.json only when the host needs a vendor-specific manifest.
  • Do not confuse Skill Steward plugin.yaml with a host marketplace manifest. It is a local meta-plugin manifest for documenting Skill Steward wiring.

Scaffold a new marketplace repo

A. Skills-only marketplace (simplest — Skill Steward style)

my-skills/
├── skills/
│   └── my-skill/
│       └── SKILL.md
├── skills.sh.json          # optional: skills.sh categories
├── README.md               # install commands
└── package.json            # optional: pnpm run validate

Publish: push public → npx skills add org/my-skills. No plugin manifest required. For customizing the repository page by defining groupings and categories in skills.sh.json, see the skills.sh customization docs.

B. Codex / Claude / Cursor marketplace (multi-plugin)

my-marketplace/
├── .agents/plugins/
│   └── marketplace.json    # Codex catalog
├── .claude-plugin/
│   └── marketplace.json    # Claude catalog, when needed
├── .cursor-plugin/
│   └── marketplace.json    # Cursor catalog, when needed
└── plugins/
    └── my-plugin/
        ├── .plugin/plugin.json
        ├── .codex-plugin/plugin.json
        ├── .claude-plugin/plugin.json
        ├── .cursor-plugin/plugin.json
        ├── skills/
        ├── hooks/
        └── .mcp.json

Codex entry: "source": {"source": "local", "path": "./plugins/my-plugin"} plus policy/category. Claude and Cursor keep their own marketplace shapes. Use only the vendor trees the target agents need.

C. Skill Steward meta-plugin (hooks + skill refs)

plugins/my-hook-pack/
├── plugin.yaml
├── README.md               # manual .cursor/hooks.json steps
├── hooks.json.snippet
└── hooks/

Canonical skills stay in skills/. See templates/plugin/.

Private marketplace checklist

  1. Repo access — team can git clone (HTTPS token, SSH, or GHE app for Cursor team marketplaces).
  2. Secrets — never commit tokens; document GITHUB_TOKEN / GITLAB_TOKEN for Claude auto-update on private marketplaces.
  3. Install docs — one README block per channel (skills CLI vs Claude /plugin vs Cursor dashboard).
  4. Versioning — bump version in plugin manifest per release; skills use git SHA via npx skills update.
  5. CI — validate skills (pnpm run validate / dart run :steward validate in Skill Steward); product repos add contract tests.
  6. Scope — project install (committed .agents/skills/) vs -g global; document team norm.

Skill Steward vs product repo

ConcernSkill StewardProduct (e.g. product MCP)
SkillsMeta only (skill-authoring-lifecycle, repository-governance-lifecycle, …)Domain + MCP skills
PluginsMeta hooks (steward-validate-on-save)Full bundle: MCP + skills + init
CLIsteward validate[toolkit-cli] init <agent>
MarketplacePublic Git + skills.shClaude/Codex git + Cursor submit + Smithery
Copy/init scriptsSkills teach the patternProduct repo owns generated script, host payloads, and install proof

Do not put product MCP servers in Skill Steward. Cross-promote: npx skills add arenukvern/skill_steward --skill mcp-harness-repo-maintainer.

Workflow: add a distributable skill to Skill Steward

  1. Follow skill-authoring-lifecycle — skills/{name}/SKILL.md.
  2. Register in skills.sh.json + root README.md.
  3. pnpm run validate in skill_steward.
  4. After merge to main, public repo is installable via npx skills add arenukvern/skill_steward --skill {name}.
  5. Optional: submit to skills.sh leaderboard (automatic for public repos with valid skills).

Workflow: add a Skill Steward plugin

  1. Copy templates/plugin/ → plugins/{id}/.
  2. List referenced skills in plugin.yaml (ids only).
  3. Document install (Cursor hooks merge, etc.).
  4. Update plugins/README.md.

Anti-patterns

  • Putting hooks only in skills/ and expecting npx skills to wire Cursor
  • Duplicating SKILL.md inside every plugin directory
  • Calling a repo-local copy/init script "public marketplace readiness" without host install proof
  • Moving product-specific installer code into steward_cli before repeated repo evidence proves a generic Steward module is useful
  • Private marketplace docs that only mention one agent
  • Eternal plan files in marketplace repos (see plan hygiene)

Additional resources

Install this skill

npx skills add arenukvern/skill_steward --skill plugin-marketplace-setup

Sources

See references/sources.md. When researching, follow skill-source-citations.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Design, implement, and integrate generalized validation harnesses across a producer-consumer boundary after a local harness contract exists. Use when refactoring custom validation CLIs/MCPs for large polyrepos, extending Steward across sibling repos, or deploying a local tool to a consumer project for dogfooding and testing; use mcp-harness-repo-maintainer first for initial steward.yaml adoption and cold-start proof.

日本語の概要は準備中です。原文の説明を表示しています。

Arenukvern/skill_steward112026年9月24日 更新

Maintains repo-local action contracts and harness repositories where product CLI and MCP adapters stay thin over core libraries. Use when adopting or improving steward.yaml actions, capability-level adoption evidence, cold-start proof loops, probes, benchmarks, CLI/MCP/core parity, adapter refactors, packages/core boundaries, or sibling harness layout; use repo-quality-system-lifecycle first for general app/library/tool stewardship baselines.

日本語の概要は準備中です。原文の説明を表示しています。

Arenukvern/skill_steward112026年9月24日 更新

Run a Mixture of Experts (MoE) audit on any topic, plan, codebase, evidence archive, or process. Dynamically spawns specialized subagents with different critical lenses to cross-reference findings and detect flaws, overlap, retention issues, or drift. Use when designing architectures, analyzing complex code, verifying multi-step plans, classifying evidence artifacts, or looking for duplicated intent in a repo.

日本語の概要は準備中です。原文の説明を表示しています。

Arenukvern/skill_steward112026年9月24日 更新

Plan and document handoffs, parent lane contracts, and parallel batch contracts between specialized AI agents (foreman, workers, reviewers). Use for multi-agent workflows, subagents, original goal preservation, native gates, claim ceilings, terminal states, baton passes, or guild-style agent coordination.

日本語の概要は準備中です。原文の説明を表示しています。

Arenukvern/skill_steward112026年9月24日 更新

Chooses ecosystem-native release and changelog tooling (Changesets, Melos, release-plz) plus binary distribution (GitHub Release tarballs, install.sh) when the product is an executable. Use for release CI, install.sh, versioning, CHANGELOGs, shipping MCP/CLI without clone, or meta repos that only ship skills via npx skills—not domain app tutorials.

日本語の概要は準備中です。原文の説明を表示しています。

Arenukvern/skill_steward112026年9月24日 更新

Establish or audit a structural quality contract for any agent-operated engineering repository: app, library, CLI/tool, plugin, harness, or meta repo. Use when a repo needs charter clarity, docs/decision ownership, type-native validation gates, evidence paths, safe action policy, cold-start legibility, maturity proof, repair-loop routing, consumer/producer owner-boundary routing, command surface evolution, evolutionary simplicity (see references/evolutionary-simplicity.md), or a repo ecology pass to decide what should be compressed, merged, removed, updated, created, retired, or moved to checks.

日本語の概要は準備中です。原文の説明を表示しています。

Arenukvern/skill_steward112026年9月24日 更新

Arenukvern のスキルをすべて見る

このスキルの問題を報告する