Claude Plugins for Development: A Guide

Overview

The development category on ClaudePluginHub holds 6,299 plugins — a sizable slice of the 33,235 plugins in the directory overall. Across the whole directory, those plugins ship 316,734 components, and skills dominate that mix at 190,406. The development category reflects the same skew: most of the plugins below extend Claude Code through skills rather than hooks or MCP servers. This guide walks through eight of them, anchored in install and star counts, so you can judge what to install rather than guess. Browse the full category at /plugins?category=development, or the entire directory at /plugins.

Why This Category Matters

"Development" is broad on purpose. It covers frontend polish, full-stack code generation, behavioral guardrails for the model itself, game engines, CMS backends, and prompt management. The common thread is that these plugins change how Claude Code writes and reviews code, not what it knows about a narrow domain. With 6,299 plugins in the category, the practical problem is no longer finding a development plugin — it is telling apart the ones with traction from the long tail. Install counts over a 7-day window and star counts give two independent signals: installs show recent pull on this directory, stars show broader GitHub interest that often predates a plugin's listing here.

Plugins to Consider

impeccable is the most-installed of this set at 272 installs in 7 days, against 29,563 stars. It ships skills that audit, critique, and polish frontend interfaces with production-grade code changes — covering UX, visual hierarchy, accessibility, performance, responsive behavior, theming, typography, color, motion, micro-interactions, and design systems. If your work is front-of-house, it is the one to look at first.

andrej-karpathy-skills carries heavy GitHub interest at 149,256 stars, with 92 installs in 7 days. Its skills reduce common LLM coding mistakes by enforcing behavioral guidelines for simplicity, surgical changes, assumption surfacing, and verifiable success criteria. It is less a feature add and more a set of guardrails on how the model behaves.

fullstack-dev-skills pairs commands with skills and posts 58 installs in 7 days against 9,269 stars. It equips Claude with full-stack expertise across multiple languages, frameworks, and cloud platforms — code generation, debugging, architecture design, DevOps automation, testing, security audits, and project management workflows. The commands component means you get explicit entry points, not just passive skills.

prompts.chat is the most-starred plugin in this set at 162,464 stars, and the broadest by component coverage — commands, agents, skills, and an MCP server. It lets you search, retrieve, improve, and manage thousands of AI prompts and Claude skills from prompts.chat directly in your assistant, fill prompt variables, and save custom prompts with metadata.

godot-skills is the honest small-fish example: 30 stars, but still 92 installs in 7 days — the same 7-day install count as the 149,256-star Karpathy plugin. It ships skills for the Godot game engine. Stars and installs do not always agree, and a low star count does not mean low recent demand.

agent-skills holds 45,461 stars and is the only plugin here spanning commands, agents, skills, and hooks. It equips coding agents with production engineering skills across full dev lifecycles: refining ideas to specs, TDD implementation, test and debug runs, multi-axis code review, performance and security optimization, CI/CD automation, and ship checklists.

slidev brings 46,588 stars and skills for building interactive Slidev slide decks from Markdown — embedding Vue components, syntax-highlighted code, animations, LaTeX equations, Mermaid diagrams, and live demos. It is the outlier here: development tooling aimed at technical talks and walkthroughs rather than shipping application code.

payload holds 42,451 stars and ships skills for configuring and debugging Payload CMS backends in payload.config.ts — defining collections, fields, hooks, access control, and APIs, and troubleshooting validation errors, relationships, queries, and transactions. If your stack already runs Payload, it is narrowly useful.

Notable Patterns

Two things stand out in the numbers. First, stars and installs are weakly correlated within this set: prompts.chat leads on stars (162,464) but reports no 7-day install figure, while impeccable leads on installs (272) on a far smaller 29,563-star base, and godot-skills matches a 149,256-star plugin on 7-day installs with just 30 stars. Treat the two as separate questions — "is this widely known?" versus "are people installing it this week?"

Second, skills carry the category. Of the eight plugins here, every one ships skills, and only three — fullstack-dev-skills, prompts.chat, and agent-skills — add commands, agents, hooks, or MCP alongside them. That mirrors the directory-wide mix, where skills (190,406) are the largest component class by a wide margin out of 316,734 total components. If you are evaluating development plugins, expect most value to arrive as skills the model loads on demand rather than as explicit commands you invoke.

Wrap-up

The development category is large — 6,299 plugins out of 33,235 in the directory — but a handful carry most of the measurable attention. Lead with installs if you want recent, directory-specific signal; lead with stars if you want a proxy for how battle-tested a project is on GitHub. The eight plugins here span frontend polish, full-stack generation, model guardrails, game tooling, CMS work, prompt management, and slide decks, so pick by the job in front of you, not by raw counts. Start from the development category and filter from there, or widen out to /plugins for the rest of the directory.