Claude Code Skills — Universal Toolkit — a reusable Claude Code Skills library you can drop into your Claude Code projects to extend agent capabilities.
Claude Code supports two types of extensions: plugins (installed via /plugin install inside Claude Code) and skills (files Claude reads to know how to behave). This page covers the universally useful ones — tools that improve any Claude Code workflow regardless of what you work on.
All entries below install via /plugin install. The only difference is which marketplace they come from, shown in the install command.
Development Workflow
Superpowers
A complete software development workflow system. Rather than jumping into code, Superpowers guides Claude through spec writing, implementation planning, and test-driven development. It triggers automatically — you do not need to invoke it manually. Once installed, Claude steps back when you describe a task, clarifies requirements, writes a plan you review, then executes it using subagents that work autonomously for extended periods.
What it includes: writing plans, TDD workflow, git worktree isolation, parallel agent dispatch, code review (sending and receiving), debugging methodology, branch completion, and skill creation.
GitHub — obra/superpowers-marketplace
/plugin install superpowers@superpowers-marketplaceDouble Shot Latte
Stops Claude from interrupting multi-step work to ask permission to continue. A separate Claude instance evaluates whether stopping makes sense. If the next step is obvious and the work is incomplete, it continues automatically. Only lets Claude stop when stopping is genuinely warranted.
GitHub — obra/superpowers-marketplace
/plugin install double-shot-latte@superpowers-marketplaceFeature Dev
A structured 7-phase workflow for building new features: discovery, codebase exploration, clarifying questions, architecture design, implementation, review, and quality check. Prevents the common failure mode of jumping straight into code without understanding the existing codebase or confirming requirements.
Invoke with /feature-dev followed by your feature description.
GitHub — anthropics/claude-plugins-official
/plugin install feature-dev@claude-plugins-officialRalph Loop
Implements autonomous iterative development loops inside your current Claude Code session. You give Claude a task and a completion signal. Claude works, tries to exit, the stop hook blocks the exit and feeds the same prompt back — with all file changes persisted between iterations. Claude reads its own previous work and improves on it until the completion signal appears. Useful for long-running build, test, and fix cycles.
Invoke with /ralph-loop "your task" --completion-promise "DONE"
GitHub — anthropics/claude-plugins-official
/plugin install ralph-loop@claude-plugins-officialCode Simplifier
Reviews recently modified code for clarity, consistency, and maintainability, then fixes what it finds. Focuses on code changed in the current session unless you instruct otherwise. Removes redundancy, improves naming, and tightens logic without changing behaviour.
GitHub — anthropics/claude-plugins-official
/plugin install code-simplifier@claude-plugins-officialSkill Creator
Creates new skills, improves existing ones, and benchmarks performance. If you find yourself repeatedly giving Claude the same instructions, this turns those instructions into a reusable, invocable skill. Also runs evals to test whether a skill triggers reliably under different phrasings.
GitHub — anthropics/claude-plugins-official
/plugin install skill-creator@claude-plugins-officialBrowser Automation
Playwright Skill
General-purpose browser automation. Claude writes custom Playwright scripts on the fly for your specific request — it is not limited to pre-built scenarios. Use it to test pages, fill forms, take screenshots, validate UX flows, check responsive design, scrape data, or automate any browser interaction. Runs with a visible browser by default so you can watch what is happening.
GitHub — lackeyjb/playwright-skill
/plugin install playwright-skill@playwright-skillDocumentation & Documentation
Context7
Pulls current, version-specific documentation for any library or framework directly into Claude’s context. Addresses a common problem: Claude’s training data has a cutoff, so advice about React, Next.js, Prisma, Django, or any actively-developed library may be outdated. Context7 fetches documentation from source repositories at query time. Use it whenever you are working with a specific library and need accurate, current API references.
Provided by Upstash. GitHub — anthropics/claude-plugins-official
/plugin install context7@claude-plugins-officialFrontend & Design
Frontend Design
Generates production-grade frontend interfaces that avoid generic AI aesthetics. Makes bold choices — distinctive typography, considered colour palettes, purposeful animations — rather than producing the same bland components every AI tool produces. Use it when you want UI code that looks like a designer touched it. Triggers automatically on frontend tasks once installed.
GitHub — anthropics/claude-plugins-official
/plugin install frontend-design@claude-plugins-officialMarp Slide
Creates professional presentation slides using Marp, a Markdown-to-slides tool. Comes with 7 pre-designed themes: default, minimal, colorful, dark, gradient, tech, and business. Handles vague requests like “make this look good” and picks an appropriate theme. Supports custom layouts, image positioning, and speaker notes. Output is a Markdown file that renders to HTML, PDF, or PowerPoint via Marp CLI.
Invoke with /marp-slide followed by your content or ask Claude to “create slides for…”
GitHub — softaworks/agent-toolkit
/plugin install marp-slide@agent-toolkitDiagrams & Visualization
Mermaid Diagrams
Creates software diagrams using Mermaid’s text-based syntax. Because diagrams are defined as text, they are version-controllable and maintainable alongside code. Covers all Mermaid diagram types: class diagrams (domain modeling, object-oriented design), sequence diagrams (application flows, API interactions), flowcharts (processes, algorithms, user journeys), entity relationship diagrams (database schemas), C4 architecture diagrams (system context, containers, components), state diagrams, git graphs, Gantt charts, and pie charts.
Triggers on requests to “diagram”, “visualize”, “map out”, or “show the flow.”
GitHub — softaworks/agent-toolkit
/plugin install mermaid-diagrams@agent-toolkitWriting & Communication
Humanizer
Removes AI writing patterns from text. Based on Wikipedia’s “Signs of AI Writing” guide, compiled from thousands of observed AI-generated texts. Detects and fixes 24 specific patterns including significance inflation, promotional language, em dash overuse, rule of three, AI vocabulary (testament, landscape, showcasing), negative parallelisms, excessive hedging, and sycophantic framing.
Use it after drafting anything that will be read by humans: blog posts, proposals, emails, documentation, LinkedIn content.
Invoke with /humanizer followed by the text to fix, or ask Claude to “humanize this text.”
GitHub — softaworks/agent-toolkit
/plugin install humanizer@agent-toolkitWriting Clearly and Concisely
Applies Strunk’s rules for clear, strong writing alongside AI pattern detection. Covers what to do (active voice, concrete nouns, cut to the bone) and what not to do (the same AI patterns the Humanizer catches). Use it when writing documentation, commit messages, error messages, UI copy, reports, or any prose that a human will read. Rewrites for force and clarity, not just detection.
GitHub — softaworks/agent-toolkit
/plugin install writing-clearly-and-concisely@agent-toolkit