Browse the full directory of Claude Code plugins — commands, agents, skills, MCP servers, and more.
Browse plugins →Eight Claude Code plugins worth evaluating, selected by install velocity, star count, and component coverage — from TDD enforcement to token-saving workflows.
The Claude Code plugin ecosystem has crossed 30,057 active plugins from 13,433 authors, with 274,158 total components spanning skills, hooks, agents, commands, MCP servers, and more. In the past week alone, 1,995 new plugins were published — up from 1,402 the week before. With 22,329 installs recorded over the last 30 days, developers are actively experimenting.
That growth creates a discovery problem. Scrolling the full plugin directory or sorting by trending surfaces popular options, but popularity alone doesn't tell you whether a plugin fits your workflow. This post highlights eight plugins worth evaluating, selected by a combination of 7-day install velocity, star count, and component coverage. Each serves a different purpose — from enforcing TDD discipline to cutting token costs.
1,753 installs/7d · 185,128 stars · Components: skills, hooks
superpowers enforces strict TDD cycles, generates detailed multi-step implementation plans, and executes them in batches or via parallel subagents. It manages isolated git worktrees for features, performs root-cause debugging and technical code reviews, and verifies tests, builds, and lints before commits or PRs. If your workflow revolves around disciplined, plan-then-execute development, this plugin adds guardrails that prevent skipping steps. The hooks integration means verification runs automatically on events — not just when you remember to invoke it.
983 installs/7d · 57,436 stars · Components: agents, skills, hooks
caveman switches Claude Code into a terse communication mode, claiming to reduce token usage by 75%. It delegates to subagents for surgical 1–2 file edits, read-only code location, and diff/PR reviews. It also generates compressed git commits, memory files, and review comments while tracking token stats. For developers watching their usage costs, caveman addresses a real pain point — long Claude Code sessions can burn through tokens quickly, and terse output doesn't sacrifice accuracy for the kind of targeted edits most sessions involve.
407 installs/7d · 178,958 stars · Components: commands, agents, skills, hooks, MCP
ecc ships 300+ agents, skills, commands, and hooks, making it the broadest plugin in this list by component coverage. It spans 5 component types — commands, agents, skills, hooks, and MCP — and targets autonomous multi-agent coding workflows across JS/TS, Python, Rust, and mobile stacks. It includes TDD enforcement, security audits, and deployment automation. The tradeoff is scope: a plugin this large may include capabilities you never use, but for teams working across multiple languages, the breadth is the point.
184 installs/7d · 26,700 stars · Components: skills
impeccable focuses on frontend development, providing skills that audit, critique, and optimize production interfaces — websites, dashboards, components, and forms. It covers UX, accessibility, performance, responsive design, theming, typography, and motion, plus anti-pattern detection. If you're shipping user-facing UI and want a second opinion on polish before merge, impeccable fills that niche. It ships only skills, keeping its footprint small and composable with other plugins.
142 installs/7d · 76,229 stars · Components: skills
ui-ux-pro-max offers 50+ UI/UX styles, color palettes, font pairings, guidelines, and charts for planning and reviewing interfaces across React, Next.js, Vue, Svelte, Tailwind, React Native, and Flutter. Where impeccable audits existing UI, ui-ux-pro-max helps you plan new interfaces — choosing a style system, pairing fonts, and reviewing design consistency across frameworks. The two complement each other if frontend quality is a priority.
105 installs/7d · 123,385 stars · Components: skills
This plugin applies Andrej Karpathy-inspired rules to guide code writing, reviews, and refactoring. It enforces simplicity, surgical changes, and verifiable success criteria to avoid common LLM coding pitfalls like overcomplication. At 123,385 stars, the community interest is notable — it reflects a real concern among developers: AI-assisted coding often trends toward over-engineered solutions, and explicit simplicity constraints push back against that tendency.
90 installs/7d · 34 stars · Components: skills
brainstorming-skill provides 30+ research-validated prompt patterns across 14 categories for product features, design, marketing, and creative problem-solving. It guides pre-implementation workflows: exploring context, clarifying requirements, proposing 2–3 options with tradeoffs, securing approval, and documenting specs. The star count is low (34) compared to the others on this list, but the install velocity (90/7d) suggests developers who find it use it actively. It addresses a gap — most plugins focus on code execution, not the structured thinking that should precede it.
87 installs/7d · 57,510 stars · Components: commands, agents
gsd automates spec-driven development: bootstrapping projects with AI-generated roadmaps and requirements, autonomously planning/executing/auditing phases via 100+ commands and agents, managing git commits, releases, PRs, backlogs, and workspaces, and generating tests, docs, and reviews. It's project-scoped rather than task-scoped — where superpowers enforces discipline on individual TDD cycles, gsd manages the full lifecycle from spec to release.
These plugins aren't interchangeable — they target different stages of the development loop and different pain points.
By workflow stage:
| Stage | Plugins |
|---|---|
| Ideation & planning | brainstorming-skill, gsd |
| Implementation | superpowers, ecc, andrej-karpathy-skills |
| Frontend polish | impeccable, ui-ux-pro-max |
| Cost optimization | caveman |
By component coverage:
ecc is the only plugin here spanning 5 component types (commands, agents, skills, hooks, MCP). superpowers and caveman each include hooks, meaning they can run automatically on events — not just when invoked. impeccable, ui-ux-pro-max, andrej-karpathy-skills, and brainstorming-skill ship skills only, which makes them lightweight and composable. gsd uses commands and agents, providing both invocable slash commands and autonomous agent behaviors.
By adoption signals:
superpowers has the highest 7-day install count in this group (1,753) and the most stars (185,128). ecc and andrej-karpathy-skills have high star counts (178,958 and 123,385 respectively) relative to their install rates, suggesting broad awareness but more selective adoption. brainstorming-skill inverts this pattern — 90 installs/7d against only 34 stars — indicating a small but active user base discovering it through the directory rather than GitHub.
Rather than installing all eight, consider what's slowing you down:
Most of these plugins are composable — you can install impeccable alongside superpowers without conflicts, since they operate on different component types and different workflow stages. The main constraint is cognitive: more plugins mean more available commands, so start with one or two that match your immediate pain point.
Browse the full directory, check trending plugins, or explore top-starred plugins to compare these against the wider ecosystem.
With 30,057 plugins in the directory and 1,995 new ones published this past week, the Claude Code ecosystem is dense enough that curation matters. The eight plugins listed here were selected by install velocity and component coverage, not editorial preference — they represent different approaches to different problems. superpowers and caveman stand out by installs; ecc stands out by component breadth; and niche plugins like brainstorming-skill and impeccable serve focused use cases that the broader plugins don't cover.
Install one, use it for a week, and evaluate whether it actually changes your workflow. That's a more reliable signal than any star count.
Enforce test-driven development (TDD) cycles, guide root-cause debugging, and collaborate through code review feedback using structured plans, parallel agent execution, and isolated git worktrees within Claude Code.
Switch to caveman mode in Claude Code sessions to slash 75% token usage with terse, accurate technical communication. Delegate to subagents for surgical 1-2 file edits, read-only code location, diff/PR reviews; generate compressed git commits, memory files, review comments, and track token stats via commands.
Supercharge Claude Code with 300+ agents, skills, commands, and hooks to orchestrate autonomous multi-agent coding workflows, enforce TDD, conduct security audits, generate production code across JS/TS/Python/Rust/mobile stacks, optimize performance, and automate deployments/testing.
Elevate frontend development with AI commands that polish, audit, critique, and optimize production-grade interfaces like websites, dashboards, components, and forms. Unlock expert reviews for UX, accessibility, performance, responsive design, theming, typography, motion, plus curated anti-pattern detection.
Access 50+ UI/UX styles, color palettes, font pairings, guidelines, and charts to plan, build, review, and optimize web and mobile interfaces across React, Next.js, Vue, Svelte, Tailwind, React Native, and Flutter.
Apply Andrej Karpathy-inspired rules to guide code writing, reviews, and refactoring: enforce simplicity, surgical changes, and verifiable success criteria to sidestep common LLM coding pitfalls like overcomplication.
Conduct structured brainstorming for product features, designs, marketing strategies, and creative problem-solving using 30+ research-validated prompt patterns across 14 categories. Guide pre-implementation workflows by exploring context, clarifying requirements, proposing 2-3 options with trade-offs, securing approval, and documenting specs.
Automate spec-driven software development in Claude Code: bootstrap projects with AI-generated roadmaps/requirements, autonomously plan/execute/audit phases via 100+ commands/agents, manage git commits/releases/PRs/backlogs/workspaces, generate tests/docs/reviews, debug/fix issues.