DEV Community

Shrijith Venkatramana
Shrijith Venkatramana

Posted on

Beyond Prompts: 10 Installable Superpowers for Coding Agents

#ai #productivity #programming #webdev

Hi there! I'm Shrijith Venkatramana, and I'm building something I think you'll find useful—git-lrc, an AI code reviewer that checks every commit before it lands in your repo. If that sounds interesting, give us a star on GitHub to help other devs discover it. And please, try it out—I'd genuinely love your feedback.

AI coding agents are becoming more capable every month. But one of the biggest shifts isn't a better model—it's teaching your agent new skills.

Instead of writing the same prompt over and over ("review my architecture", "design a better UI", "help me write tests"), you can install reusable skills that encode proven workflows and best practices.

The open skills.sh ecosystem has already crossed 1 million+ installs, with hundreds of reusable skills available for Claude Code, Cursor, GitHub Copilot, Cline, and other compatible agents. :contentReference[oaicite:0]{index=0}

Let's look at the current Top 10 skills, what each one does, and when you should actually use it.

Installing Skills

First, install the Skills CLI.

npx skills add <owner/repository>
Enter fullscreen mode Exit fullscreen mode

If the repository contains multiple skills:

npx skills add <owner/repository> --skill <skill-name>
Enter fullscreen mode Exit fullscreen mode

Useful commands:

# Search for skills
npx skills find react

# Check for updates
npx skills check

# Update installed skills
npx skills update
Enter fullscreen mode Exit fullscreen mode

The CLI works with many popular AI coding agents including Claude Code, Cursor, GitHub Copilot, and Cline. :contentReference[oaicite:1]{index=1}

1. find-skills (vercel-labs/skills)

What it does

Think of this as the package manager for AI skills.

Whenever your agent doesn't know the best workflow for something, this skill teaches it how to search the Skills ecosystem and recommend high-quality skills.

Install:

npx skills add vercel-labs/skills --skill find-skills
Enter fullscreen mode Exit fullscreen mode

Use it by asking:

"Find me a skill for Docker deployments."

or

"Is there a skill for writing better documentation?"

It searches the ecosystem, recommends trustworthy skills, and even suggests install commands.

Best for

  • Discovering new capabilities
  • Exploring the ecosystem
  • Finding community skills

2. frontend-design (anthropics/skills)

What it does

Makes your AI much better at UI and frontend design.

Instead of generating generic HTML, it starts thinking more like an experienced product designer by considering:

  • visual hierarchy
  • spacing
  • typography
  • color usage
  • responsive layouts
  • accessibility
  • modern UI patterns

Install:

npx skills add anthropics/skills --skill frontend-design
Enter fullscreen mode Exit fullscreen mode

Try prompts like:

"Redesign this dashboard."

"Make this login page feel premium."

"Improve the visual hierarchy."

You'll usually get noticeably better design suggestions than with a vanilla model.

Best for

  • React apps
  • SaaS dashboards
  • Landing pages
  • Internal tools

3. grill-me (mattpocock/skills)

What it does

This is one of my favorites.

Instead of agreeing with everything you say, the AI becomes a skeptical reviewer.

It starts asking difficult questions like:

  • Why?
  • Have you considered X?
  • What if this fails?
  • What assumptions are you making?

Install:

npx skills add mattpocock/skills --skill grill-me
Enter fullscreen mode Exit fullscreen mode

Example prompt:

"Grill my API design."

or

"Challenge my startup idea."

It feels like discussing your work with a strong senior engineer.

Best for

  • Architecture reviews
  • Design reviews
  • Startup ideas
  • Technical proposals

4. agent-browser (vercel-labs/agent-browser)

What it does

Teaches your agent browser automation workflows.

Instead of just talking about web pages, the agent can reason about interacting with websites and browser-driven tasks.

Install:

npx skills add vercel-labs/agent-browser
Enter fullscreen mode Exit fullscreen mode

Example requests:

"Test this signup flow."

"Navigate the application and summarize the UX."

"Check whether this responsive layout breaks."

Best for

  • QA
  • Browser automation
  • End-to-end testing
  • UI inspection

5. vercel-react-best-practices

What it does

A huge collection of React and Next.js best practices from Vercel.

It teaches your AI things like:

  • rendering patterns
  • performance
  • component design
  • React Server Components
  • data fetching
  • caching
  • composition
  • maintainability

Install:

npx skills add vercel-labs/agent-skills --skill vercel-react-best-practices
Enter fullscreen mode Exit fullscreen mode

Now ask:

"Refactor this React component."

or

"Improve the performance of this page."

The suggestions become much closer to what experienced React developers would recommend.

Best for

  • React
  • Next.js
  • Performance optimization
  • Code reviews

6. grill-with-docs (mattpocock/skills)

What it does

Similar to grill-me, but instead of challenging your ideas from experience alone, it uses documentation as evidence.

Rather than saying:

"I think this is wrong."

it says:

"According to the documentation, this pattern isn't recommended."

Install:

npx skills add mattpocock/skills --skill grill-with-docs
Enter fullscreen mode Exit fullscreen mode

Great prompts:

"Review this API usage."

"Challenge this implementation against the framework docs."

Best for

  • Learning frameworks
  • Avoiding outdated APIs
  • Documentation-driven reviews

7. improve-codebase-architecture

What it does

Looks beyond individual files.

Instead of suggesting variable renames, it examines:

  • project structure
  • module boundaries
  • coupling
  • abstractions
  • maintainability
  • scalability

Install:

npx skills add mattpocock/skills --skill improve-codebase-architecture
Enter fullscreen mode Exit fullscreen mode

Example:

"Review this repository's architecture."

or

"How would you reorganize this project?"

This is particularly useful on mature codebases.

Best for

  • Refactoring
  • Large repositories
  • Technical debt
  • Long-term maintainability

8. tdd (mattpocock/skills)

What it does

Encourages the agent to work using Test-Driven Development.

Instead of immediately generating implementation code, it tends to:

  1. write tests
  2. think through edge cases
  3. implement only what's needed
  4. refactor afterwards

Install:

npx skills add mattpocock/skills --skill tdd
Enter fullscreen mode Exit fullscreen mode

Example prompt:

"Build a URL parser using TDD."

or

"Let's implement this feature test-first."

Excellent if you're trying to build more reliable software.

Best for

  • Library development
  • Backend code
  • Complex business logic
  • Learning TDD

9. web-design-guidelines

What it does

Focuses specifically on good web design principles.

Unlike frontend-design, which improves overall UI generation, this skill emphasizes:

  • usability
  • accessibility
  • spacing
  • responsive behavior
  • layout consistency
  • navigation
  • readability

Install:

npx skills add vercel-labs/agent-skills --skill web-design-guidelines
Enter fullscreen mode Exit fullscreen mode

Try:

"Audit my landing page."

"Improve accessibility."

"Review this homepage."

It's particularly useful before shipping a public website.

Best for

  • Marketing sites
  • Documentation
  • Landing pages
  • Design reviews

10. microsoft-foundry

What it does

Provides guidance for building AI applications using Microsoft's AI Foundry ecosystem.

If you're working with Azure AI services, this skill helps the agent recommend the right workflows, architectures, and platform capabilities.

Install:

npx skills add microsoft/azure-skills --skill microsoft-foundry
Enter fullscreen mode Exit fullscreen mode

Example prompts:

"Help me build an AI application on Azure."

"Recommend an AI Foundry architecture."

"How should I deploy this AI workload?"

Best for

  • Azure users
  • Enterprise AI
  • Microsoft AI services

Which Skills Should You Install First?

If you're a general software engineer, I'd start with these five:

Skill Why install it?
find-skills Discover everything else
grill-me Better technical decisions
improve-codebase-architecture Higher-quality code reviews
tdd More reliable implementations
vercel-react-best-practices Essential if you write React

Frontend developers should also add:

  • frontend-design
  • web-design-guidelines

Azure users should definitely install:

  • microsoft-foundry

Skills Are Becoming the New Prompt Library

One thing stands out after looking through the leaderboard: these skills aren't replacing AI models—they're teaching them repeatable workflows.

Instead of repeatedly explaining how you want your AI to think, you install that expertise once and reuse it across every conversation.

As the ecosystem grows, I expect most developers to have a personal toolkit of skills for architecture, testing, design, documentation, DevOps, and domain-specific workflows—much like we already rely on packages and libraries today.

Which AI skill has saved you the most time? Have you published one yourself? I'd love to hear what's in your toolbox.

AI agents write code fast. They also silently remove logic, change behavior, and introduce bugs—without telling you. You often find out in production.

git-lrc fixes this. It hooks into git commit and reviews every diff before it lands. 60-second setup. Completely free.

Any feedback or contributors are welcome! It's online, source-available, and ready for anyone to use.

GitHub logo HexmosTech / git-lrc

Free, Micro AI Code Reviews That Run on Git Commit



GenAI today is a race car without brakes. It accelerates fast -- you describe something, and large blocks of code appear instantly. But AI agents silently break things: they remove logic, relax constraints, introduce expensive cloud calls, leak credentials, and change behavior -- without telling you. You often find out in production.

git-lrc is your braking system. It hooks into git commit and runs an AI review on every diff before it lands. 60-second setup. Completely free.

In short, git-lrc helps Prevent Outages, Breaches, and Technical Debt Before They Happen

At a glance: 10 risk categories · 100+ failure patterns tracked · every commit…

View on GitHub

Top comments (0)