Awesome Claude Code Toolkit: 135 Agents (2026)

The awesome-claude-code-toolkit repository by Rohit Ghumare (1.4K+ stars) is a specialized catalog of Claude Code agents, skills, commands, plugins, and hooks. It tracks 135 agents, 35 skills (plus 400K+ via SkillKit integration), 42 commands, 176+ plugins, and 20 hooks — all with compatibility notes and install instructions.

What It Is

A curated index that goes deeper than general-purpose awesome lists. Where awesome-claude-code catalogs the full ecosystem, this toolkit focuses specifically on actionable components you install into Claude Code:

  • 135 Agents — pre-configured behavioral profiles for specialized roles
  • 35 Core Skills — standalone CLAUDE.md rule packages
  • 400K+ SkillKit Skills — integration with the SkillKit marketplace
  • 42 Commands — custom slash commands for .claude/commands/
  • 176+ Plugins — full feature packages combining skills, commands, and hooks
  • 20 Hooks — lifecycle scripts for session start, pre/post tool execution

Each entry includes install method, compatibility version, description, and a link to the source repo.

Why It Matters

Finding Claude Code components is scattered across GitHub, npm, and community forums. This toolkit consolidates them into one searchable, categorized list. The focus on install-ready components (not tutorials or articles) means every entry is something you can use immediately.

The SkillKit integration is notable — it bridges the gap between the 35 manually curated skills and the 400K+ skills available in the SkillKit marketplace, with quality ratings to help you filter.

Installation

The toolkit itself is a reference list. Browse it on GitHub:

open https://github.com/rohitg00/awesome-claude-code-toolkit

To install listed components, follow each entry’s install instructions. Common patterns:

# Agent (CLAUDE.md behavioral profile)
curl -o .claude/agents/security-auditor.md \
  https://raw.githubusercontent.com/example/agents/security-auditor.md
# Command (slash command)
curl -o .claude/commands/review.md \
  https://raw.githubusercontent.com/example/commands/review.md
# Hook (lifecycle script)
curl -o .claude/hooks/pre-commit-lint.sh \
  https://raw.githubusercontent.com/example/hooks/pre-commit-lint.sh
chmod +x .claude/hooks/pre-commit-lint.sh

Key Features

  1. 135 Categorized Agents — agents grouped by role: code reviewer, test writer, documentation generator, security auditor, DevOps engineer, database architect, API designer, and more.

  2. Quality Tiers — entries rated by community usage and maintenance status. Top-tier agents have 1K+ installs and active maintenance.

  3. SkillKit Bridge — connects to the SkillKit marketplace for 400K+ additional skills. Quality ratings help filter the signal from the noise.

  4. Install Commands — every entry includes the exact command to install. No hunting through READMEs.

  5. Compatibility Matrix — entries tagged with supported Claude Code versions. Avoid installing plugins that require features your version lacks.

  6. Plugin Bundles — the 176+ plugins are pre-assembled combinations of skills, commands, and hooks that work together. Install one plugin, get a coherent toolset.

For more on this topic, see Claude Code for StyleX Meta.

For more on this topic, see Claude Code for AeroSpace Tiling.

For more on this topic, see Claude Code for Particle Physics ROOT.

  1. Hook Library — 20 curated hooks for common needs: linting, testing, logging, cost tracking, and documentation generation.

  2. Weekly Updates — the maintainer adds new entries multiple times per week as the ecosystem grows.

Real Usage Example

Finding and Installing an Agent

Need a code review agent? Browse the Agents category:

Category: Code Review
  1. strict-reviewer (★★★★★) — 5-severity scale, OWASP-aware
  2. gentle-reviewer (★★★★☆) — educational tone, suggestions not demands
  3. perf-reviewer (★★★★☆) — performance-focused, benchmarks included
  4. security-reviewer (★★★★★) — OWASP Top 10, CVE checks, secret scanning

Install strict-reviewer:

mkdir -p .claude/agents
curl -o .claude/agents/strict-reviewer.md \
  https://raw.githubusercontent.com/example/strict-reviewer.md

Reference it in CLAUDE.md:

# Code Review Mode
When reviewing code, follow the behavioral rules in .claude/agents/strict-reviewer.md

Installing a Plugin Bundle

The “full-stack-quality” plugin bundles a reviewer agent, test-writer command, and lint hook:

npx claude-toolkit install full-stack-quality

This creates:

  • .claude/agents/quality-reviewer.md
  • .claude/commands/write-tests.md
  • .claude/hooks/pre-commit-lint.sh
  • Updates .claude/settings.json with hook registrations

Browsing via SkillKit

# Search SkillKit marketplace through the toolkit
npx claude-toolkit search "react testing"
Results:
  1. react-test-agent (4.8★, 12K installs) — generates RTL tests
  2. react-e2e-agent (4.5★, 8K installs) — Playwright test generation
  3. react-snapshot-agent (4.2★, 3K installs) — snapshot test management

When To Use

  • Building a Claude Code toolkit — browse and install components à la carte
  • Evaluating agent quality — the rating system helps you pick the best option
  • Discovering hooks — the 20 curated hooks cover the most common automation needs
  • SkillKit exploration — the bridge to 400K+ skills saves you from browsing the marketplace cold
  • Team standardization — agree on which agents, commands, and hooks to use across projects

When NOT To Use

  • Looking for tutorials — this is a component catalog, not a learning resource; use claude-howto instead
  • Need a framework — for structured workflows with modes and orchestration, use SuperClaude
  • Minimal setups — if you need 2-3 rules, hand-write them in CLAUDE.md

FAQ

How are quality ratings determined?

Install count, GitHub stars, maintenance frequency, and community reviews. The maintainer also manually tests top-rated entries.

Can I submit my own agent/plugin?

Yes. Open a PR with the component metadata (name, description, install URL, category, compatibility). The maintainer reviews within a week.

Do plugins auto-update?

No. Installed components are local files. Re-install to get updates, or use a hook that checks for new versions on session start.

What’s the difference between an agent and a skill?

Agents are behavioral profiles (how Claude Code should act). Skills are rule sets (what Claude Code should and shouldn’t do). In practice, the line is blurry — many entries function as both.

Our Take

7/10. Solid catalog for Claude Code component discovery. The 135 agents and 176 plugins are well-organized, and the SkillKit bridge is genuinely useful. Loses points for lower star count (1.4K vs. 40K for awesome-claude-code), which means the community validation is thinner. Some entries feel like padding — minor variations of the same agent template. Best used alongside the main awesome list for a complete picture.

See Also

Try it: Browse 155+ skills in our Skill Finder.