| name | skill-maintainer |
| description | Guides versioning, validating, releasing, and promoting plugins and skills under the agent-skills Claude-first marketplace model — plus marketplace.json / plugin.json manifest and registry alignment. Use when cutting a release, bumping a version, editing the .claude-plugin/marketplace.json catalog or a plugin.json manifest, aligning the registry, or updating the governance/validation/release tooling. For writing or scaffolding a skill use skill-studio:skill-create; for auditing skill quality use skill-studio:skill-audit. |
Skill & Plugin Maintainer
Meta-tooling for maintaining the agent-skills marketplace itself — the
Claude Code plugin marketplace. Use it when changing how the marketplace
packages, validates, versions, or promotes its own plugins and skills. For the
governance model it enforces, see docs/marketplace-governance.md.
Apply When
- creating or changing a skill inside a plugin (
plugins/<plugin>/skills/<skill>/)
- adding or promoting a plugin, or editing
.claude-plugin/marketplace.json
- editing a
plugin.json manifest (name, version, author, description)
- reconciling a skill package:
SKILL.md frontmatter, metadata.json, CHANGELOG.md
- updating the governance docs,
scripts/validate-skill.sh, or release tooling
- deciding whether an experimental skill is ready to graduate from a sandbox
plugin to an official one
Do Not Apply When
- the task is to USE an existing skill to solve an app or platform problem
(use that skill directly)
- the work is authoring a plugin/skill outside the agent-skills repo
- the task is only committing already-complete changes
- the change edits source contracts without changing marketplace guidance
Routing Table
Procedure
- Classify the request: new skill, skill update, version-only, plugin/manifest,
marketplace, governance-doc, or tooling work.
- Decide the right tier and home: an official plugin (e.g.
repo-governance)
vs a personal sandbox plugin (e.g. blassioli). See
references/marketplace-maintenance.md.
- Define the skill boundary in one sentence (triggers + anti-triggers).
- Read the relevant source contracts before editing guidance.
- Update the skill package together:
SKILL.md (frontmatter = name +
description only), metadata.json (version/date/source_contracts),
CHANGELOG.md, optional README.md, references, assets.
- Update
.claude-plugin/marketplace.json and the plugin.json version for new
plugins or releasable changes.
- Keep cross-skill references by name (not relative filesystem links); keep
bundled files inside the skill and reference them via
${CLAUDE_SKILL_DIR} /
${CLAUDE_PLUGIN_ROOT}.
- Validate before sign-off (see below).
Confirmation Policy
- Confirm before promoting a skill from a sandbox plugin to an official one.
- Confirm before changing the marketplace schema, a plugin name (= namespace), or
install semantics.
- Confirm before splitting one requested skill into multiple siblings.
- Confirm before removing source contracts or narrowing activation triggers.
- Confirm before changing validator behavior or the release workflow.
Validation
claude plugin validate ./plugins/<plugin> --strict
claude plugin validate . --strict
bash scripts/validate-skill.sh plugins/<plugin>/skills/<skill>
bash scripts/test-validate-skill.sh