npx claudepluginhub alexanderop/claude-code-builderCreate skills, subagents, hooks, commands, and plugins for Claude Code
Build skills, subagents, hooks, slash commands, and plugins for Claude Code — fast, consistently, and with best practices baked in.
Creating high-quality Claude Code customizations involves repetitive scaffolding and convention wrangling. Claude Code Builder gives you guided commands that generate the right files, structure, and docs — so you can focus on logic, not boilerplate.
For local development, contributors open a PR against this repo.
When the PR is merged into main, the marketplace manifest already points to this repo, so the plugin is live immediately (no extra publish step). Reviews and approvals are handled by the maintainer.
Summary
main → live in marketplace1) Register the marketplace in Claude Code:
/plugin marketplace add alexanderop/claude-code-builder
2) Install the plugin from this marketplace:
/plugin install claude-code-builder@claude-code-builder
Check that commands appear:
/help
# Should see:
# /create-skill - Create an auto-invoked Skill with best practices
# /create-agent - Create a subagent with focused role/tools
# /create-command - Generate a slash command (explicit invocation)
# /create-hook - Configure event-driven hooks (pre/post tool use, etc.)
# /create-plugin - Package a .claude/ setup into a plugin
# /create-md - Generate CLAUDE.md files (project awareness)
# /create-output-style - Create/activate custom Output Styles
Tip: If commands don’t appear, run
/helpagain or restart Claude Code, then re-run the install command.
| Command | Purpose (1-liner) | Typical usage example |
|---|---|---|
/create-skill | Scaffold a model-invoked Skill Claude auto-discovers from context. | /create-skill commit-helper "Generate helpful git commit messages" |
/create-agent | Create a subagent with a focused role and optional tool/model settings. | /create-agent reviewer "Expert code reviewer; use proactively after changes" |
/create-command | Add a slash command for repeatable workflows you invoke explicitly. | /create-command analyze-deps "Analyze dependencies for outdated packages" |
/create-hook | Configure event hooks (pre/post tool use, prompt submit, etc.) with JSON/CLI. | /create-hook PreToolUse "Bash" "jq -r '.tool_input.command' >> ~/.log" |
/create-plugin | Turn a .claude/ setup into a distributable plugin with manifests. | /create-plugin |
/create-md | Generate root & nested CLAUDE.md files that teach Claude your project. | /create-md medium |
/create-output-style | Create/activate Output Styles (stored system prompts) for different modes. | /create-output-style "Teaching Assistant" --project --description "..." |
Each command opens a guided flow and prints the exact files/paths it creates.
# Skill (auto-invoked by Claude)
#/create-skill [skill-name] "[what it does and when to use it]"
/create-skill commit-helper "Generate clear commit messages; use when committing or reviewing staged changes."
# Subagent (focused role)
#/create-agent [name] "[when to invoke]" [--tools ...] [--model ...]
/create-agent reviewer "Expert code reviewer; use proactively after code changes" --tools "Read,Grep,Glob"
# Slash command (explicit invocation)
/create-command analyze-deps "Analyze dependencies for outdated packages"
# Hook (automate on events)
/create-hook PreToolUse "Edit|Write" "python3 .scripts/block_sensitive_edits.py"
# Package your setup as a plugin
/create-plugin
# Project awareness docs for Claude
/create-md very-thorough
# Output style for non-default workflows
/create-output-style "Teaching Assistant" --project --description "Explains step by step with examples"