npx claudepluginhub javimontano/jm-adk-alfaThis skill uses the workspace's default tool permissions.
agents/guardian.mdagents/lead.mdagents/specialist.mdagents/support.mdevals/evals.jsonknowledge/body-of-knowledge.mdknowledge/knowledge-graph.mdprompts/meta.mdprompts/primary.mdprompts/variations/deep.mdprompts/variations/quick.mdreferences/domain-knowledge.mdtemplates/output.docx.mdtemplates/output.htmlSearches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
--- [EXPLICIT] name: plugin-builder argument-hint: "plugin-name" [EXPLICIT] description: Builds complete Claude Code plugins from scratch — scaffolds plugin.json, CLAUDE.md, commands/, agents/, skills/, hooks/, scripts/, settings.json. Activates when the user says "create a plugin", "scaffold a plugin", "build a new plugin", "initialize plugin structure", or "set up a plugin project". Also triggers on mentions of plugin creation, plugin scaffolding, or Claude Code extension packaging. Use this skill even if the user only names the plugin — it interviews for missing details. [EXPLICIT] model: opus context: fork allowed-tools:
Creates production-ready Claude Code plugins following the official plugin specification. [EXPLICIT]
{plugin-name}/
├── .claude-plugin/
│ └── plugin.json ← Manifest (only file here)
├── commands/ ← User-invocable skills (/plugin:command)
│ └── {verb}-{context}.md ← Verb-first naming
├── agents/ ← Custom agent definitions
│ └── {role-name}.md ← Role-based naming
├── skills/ ← Agent skills (model-invoked)
│ └── {skill-name}/
│ └── SKILL.md ← Frontmatter + instructions
├── hooks/
│ └── hooks.json ← Event handlers
├── scripts/ ← Shell scripts for hooks
├── settings.json ← Default settings (e.g., default agent)
├── CLAUDE.md ← Orchestration guide
└── README.md ← User documentation
{
"name": "{kebab-case-name}",
"version": "1.0.0",
"description": "{1-2 sentence purpose}",
"author": {
"name": "{Author Name}",
"url": "{homepage}"
},
"homepage": "{url}",
"repository": "{git-url}",
"license": "SEE LICENSE IN LICENSE",
"keywords": ["{keyword1}", "{keyword2}"]
}
Critical: name field = command prefix. Users type /name:command. Keep it SHORT.
---
description: "{What it does — shown in /help}"
user-invocable: true
---
# {TITLE}
{System prompt for this command. This is what Claude reads when the user invokes it.}
## ROLE
{Who Claude becomes}
## PROTOCOL
{Step-by-step instructions}
## OUTPUT CONFIGURATION
- Language: {target language}
- Format: {markdown/html/etc}
## CONSTRAINTS
{What NOT to do}
---
name: plugin-builder
description: "{Role and when activated}"
co-authored-by: {Author} (with Claude Code)
---
# {Role Title}
{You are a... + responsibilities + skills assigned}
---
name: plugin-builder
author: {Team/Author}
description: >
{When to use this skill — trigger phrases and keywords}
model: opus
context: fork
allowed-tools:
- Read
- Write
- Edit
- Glob
- Grep
- Bash
---
# {Skill Title}
{Purpose statement}
## Guiding Principle
> *{Philosophy in 1-3 sentences}*
## Procedure
{Step-by-step methodology}
## Quality Criteria
{Acceptance criteria}
## Anti-Patterns
{What NOT to do}
{
"hooks": {
"SessionStart": [
{
"matcher": "startup",
"hooks": [
{"type": "command", "command": "${CLAUDE_PLUGIN_ROOT}/scripts/startup.sh", "timeout": 10}
]
}
],
"PostToolUse": [
{
"matcher": "Write|Edit",
"hooks": [
{"type": "command", "command": "${CLAUDE_PLUGIN_ROOT}/scripts/validate.sh", "timeout": 15}
]
}
]
}
}
---
description: "Alias → {target-command}. Use /{plugin}:{target-command} instead."
user-invocable: true
---
# Redirect: {alias} → {target-command}
Activate `{target-command}` skill with all arguments passed through: $ARGUMENTS [EXPLICIT]
Each skill can be packaged as a ZIP with extension .skill: [EXPLICIT]
sofka-{name}.skill (ZIP containing):
sofka-{name}/
SKILL.md
references/
body-of-knowledge.md
state-of-the-art.md
knowledge-graph.mmd
prompts/
metaprompts.md
use-case-prompts.md
examples/
sample-output.md
sample-output.html
The entire plugin can be packaged as a ZIP with extension .plugin: [EXPLICIT]
{name}.plugin (ZIP containing):
{name}/
plugin.json
settings.json
CLAUDE.md
README.md
agents/*.md
commands/*.md
hooks/hooks.json
scripts/*.sh
claude --plugin-dir ./{name} to verifyExample invocations: [EXPLICIT]
| Scenario | Handling |
|---|---|
| Empty or minimal input | Request clarification before proceeding |
| Conflicting requirements | Flag conflicts explicitly, propose resolution |
| Out-of-scope request | Redirect to appropriate skill or escalate |