Awesome Claude Skills

A curated collection of Claude AI skills, agents, and tools to supercharge your AI-powered development workflow. This repository features production-ready skills for coding, security, marketing, and specialized domains.

Why Awesome Claude Skills?

Claude skills are modular capabilities that extend Claude's functionality for specific tasks. Think of them as plugins or specialized knowledge modules that help Claude become an expert in particular domains. Whether you're building software, analyzing security vulnerabilities, or creating marketing content, there's a skill to enhance your workflow.

This list focuses on verified, actively maintained projects that provide real value to developers and teams working with Claude AI.

Contents

• Getting Started
• Official Resources
• Comprehensive Skill Collections
• Development & Engineering
• AI Slop Elimination
• Multi-Agent Systems
• Security & Compliance
• Marketing & Content
• Domain-Specific Skills
• Productivity Tools
• Learning & Documentation
• Skill Development

Getting Started

What are Claude Skills?

Claude skills are structured instructions and configurations that guide Claude to perform specialized tasks. They typically include:

• Context and expertise: Domain-specific knowledge and best practices
• Workflows: Step-by-step processes for complex tasks
• Templates: Reusable patterns for common operations
• Tool integrations: Connections to external services and APIs

Official Resources

Start here for official tools and documentation from Anthropic.

• anthropics/claude-code
• Official Claude Code agentic coding tool
• Lives in your terminal and understands your codebase
• AI-powered development assistant
• Built by Anthropic

• anthropics/skills
• Official repository for Claude Agent Skills
• Includes document processing (docx, pdf, pptx, xlsx)
• Art and design capabilities
• MCP (Model Context Protocol) builder

• anthropics/claude-code-action
• GitHub Action for Claude Code integration
• Automated code review and analysis
• CI/CD pipeline integration
• Official Anthropic GitHub Actions support

• anthropics/claude-plugins-official
• Official Anthropic-managed plugin directory
• Curated collection of high-quality Claude Code plugins
• Verified and maintained by Anthropic
• Includes internal and external plugins

• anthropics/claude-code-security-review
• AI-powered security review GitHub Action
• Analyze code changes for security vulnerabilities
• Context-aware security analysis for pull requests
• Deep semantic security analysis using Claude
• Official Anthropic security tooling

• anthropics/claude-code-base-action
• Base GitHub Action for custom Claude Code workflows
• Build any custom workflow on top of Claude Code
• Settings and MCP configuration support
• Cloud provider integration (AWS Bedrock, GCP Vertex AI)
• Official foundation for Claude Code automation

• anthropics/claude-for-legal
• Official Anthropic suite of plugins for legal workflows
• Targets common legal-team tasks across drafting, review, and search
• Python-based; maintained by Anthropic

Comprehensive Skill Collections

These repositories offer extensive collections of skills across multiple domains. Great starting points if you want a broad toolkit.

• rohitg00/awesome-claude-code-toolkit
• Most comprehensive toolkit available
• 135 agents, 35 skills, 42 commands
• 150+ plugins, 19 hooks, 7 templates
• Covers development, DevOps, and productivity

• alirezarezvani/claude-skills
• 192+ professional skills and agent plugins
• Engineering, marketing, product management
• Compliance and C-level advisory capabilities
• Compatible with 11 different tools

• JasonColapietro/suede-creator-skills
• 67 MIT-licensed skills for Claude Code and Codex
• Covers code review and ship grading, design, marketing, agent workflows, and mobile release preparation
• Installable through the Claude and Codex plugin marketplaces or the skills CLI

• Claude Code Skills 中文精选集
• Chinese directory for Claude Code Skills, agents, and plugins
• 140+ curated resources for Chinese-speaking developers
• Useful for discovering practical skills by category

• Remote OpenClaw Claude Skills
• Claude-focused skills directory with reviewed GitHub listings, plugins, and agents
• Includes adjacent Codex, Hermes, and OpenClaw hubs for cross-agent comparison
• Useful when you want a Claude-specific entry point without losing the broader ecosystem view

• CreatorSkills
• Curated marketplace of 30+ downloadable AI skills for content creators
• Covers YouTube scripting, sponsorship analysis, content repurposing, and audience growth
• Skills use the open SKILL.md format and work with Claude, ChatGPT, and 20+ AI platforms
• Paid marketplace; skills are domain-specific and ready to install

• garrytan/gstack
• Garry Tan's exact Claude Code setup
• 6 opinionated tools for startup leadership
• Includes CEO, Engineering Manager, Release Manager, QA agents
• Production-tested by Y Combinator president

• wshobson/agents
• 112 specialized agents across 72 focused plugins
• 16 orchestrators for complex workflows
• 146 skills and 79 tools
• Well-organized by use case

• swarmclawai/andrej-karpathy-skills
• Npm-installable Karpathy-inspired coding-agent guidelines
• Adapters for Claude Code, Codex, Cursor, Gemini CLI, OpenCode, OpenClaw, Windsurf, Aider, and more
• Installs with npx @swarmclawai/andrej-karpathy-skills --agent claude --dest .

• VoltAgent/awesome-agent-skills
• 1000+ agent skills from official dev teams
• Contributions from Sentry, Trail of Bits, Expo, DuckDB, Figma, Google
• Community-driven with regular updates

• sickn33/agentic-awesome-skills
• 1,326+ installable agentic skills
• CLI installer for easy setup
• Bundled workflows for common tasks

• SuperClaude-Org/SuperClaude_Framework
• Configuration framework for structured development
• 30 slash commands covering complete development lifecycle
• Cognitive personas and behavioral modes
• Deep research capabilities with autonomous web research
• MCP server integration and workflow automation

• hesreallyhim/awesome-claude-code
• One of the largest community-curated Claude Code lists
• Indexes skills, hooks, slash commands, agent orchestrators, applications, and plugins
• Tagged for agentic coding and AI workflow optimization
• Frequently referenced as a starting point for Claude Code newcomers

• ComposioHQ/awesome-claude-skills
• Composio-curated directory of Claude Skills, plugins, and workflow integrations
• Covers agent skills, MCP servers, Claude Code, Codex, Cursor, Gemini CLI, and Antigravity
• Sections for SaaS automation, developer tools, and Rube integrations
• Active community contributions with Python-driven generator scripts

• travisvn/awesome-claude-skills
• Curated list focused on Claude Skills with a Claude Code emphasis
• Includes Claude Code, Claude Desktop, and Claude AI tooling
• Lightweight directory designed for fast scanning

• muratcankoylan/Agent-Skills-for-Context-Engineering
• Open collection of agent skills focused on context engineering principles
• Foundational, architectural, operational, and methodology skill tracks
• Skills cover context compression, degradation patterns, memory systems, tool design, and LLM-as-judge evaluation
• Designed for production-grade agent systems on any agent platform

• VoltAgent/awesome-openclaw-skills
• 5,400+ skills filtered and categorized from the OpenClaw Skills Registry
• Indexed by domain with quality and recency filters
• Companion to VoltAgent's awesome-agent-skills and awesome-claude-code-subagents lists

• huggingface/skills
• Skills for plugging the Hugging Face ecosystem into AI coding agents
• Covers model search, dataset access, Spaces, and Inference Endpoints
• Python-driven, maintained by Hugging Face

• microsoft/skills
• Microsoft-published skills, MCP servers, and Custom Agents
• Ships AGENTS.md bundles for grounding coding agents
• TypeScript-driven; useful starting point for enterprise teams

• antfu/skills
• Anthony Fu's curated collection of agent skills
• Skills cover web tooling, build configuration, and developer ergonomics
• TypeScript implementation

• ConardLi/garden-skills
• ConardLi's open-source skills collection
• Web design, knowledge retrieval, and image generation skills
• CSS-driven visual assets

• heilcheng/awesome-agent-skills
• Tutorials, guides, and agent-skills directories in one place
• Aggregates resources across the agent-skills ecosystem
• TypeScript-based site for browsing

• libukai/awesome-agent-skills
• Chinese-language ultimate guide to Agent Skills
• Quick start, resource recommendations, curated skills, and tooling
• Python-driven generator

• langgptai/awesome-claude-prompts
• Curated Claude prompts for everyday use cases
• Helps non-developers get more from Claude
• Maintained by LangGPT

• davepoon/buildwithclaude
• Hub for Claude Skills, Agents, Commands, Hooks, Plugins, and Marketplace collections
• Extends Claude Code and Claude Desktop
• Python-driven discovery layer

• jnMetaCode/superpowers-zh
• Chinese expansion of obra/superpowers
• Translates the upstream skill bundle and adds six original Chinese skills
• Shell-based installer

• mergisi/awesome-openclaw-agents
• 162 production-ready AI agent templates with SOUL.md configs
• Templates organized into 19 categories
• HTML-driven gallery

• clawdbot-ai/awesome-openclaw-skills-zh
• Chinese-language OpenClaw skill library
• Translated and reorganized by scenario for natural-language invocation
• Companion to the official Clawdbot skills directory

• connerkward/ckw-skills
• Human-in-the-loop visual studios and deterministic design tooling by Conner K. Ward
• lookdev / lookdev-auto (tune AI output by eye or via a vision model), deterministic-design, ckw-design
• screenstudio-alt and macOS system-audio screen recorder, web-media-getter
• Install: /plugin marketplace add connerkward/ckw-skills
• mailtrap/mailtrap-skills
• Agent skills for Mailtrap email sending, sandbox testing, and contact management
• Covers email sending, sandbox testing, templates, domain setup, and contacts
• Follows the Agent Skills open standard (Anthropic, Dec 2025)

• abubakarsiddik31/claude-skills-collection
• Curated collection of official and community Claude Skills
• Spans productivity, creativity, coding, and content workflows
• Actively maintained general-purpose skills library

• karanb192/awesome-claude-skills
• Curated "awesome" list of 50+ verified skills
• Covers Claude Code, Claude.ai, and the API
• Document processing, git workflows, TDD, and debugging; MIT licensed

• borghei/Claude-Skills
• 368 skills, 76 expert agents, and 859 standard-library Python tools
• Spans engineering, PM, marketing, C-level, compliance, and research
• Persona system (startup-cto, growth-marketer, etc.); works with Claude Code, Cursor, Codex, Gemini, and Copilot

• levnikolaevich/claude-code-skills
• Standalone engineering skills organized by numbered categories
• Review, audit, optimization, testing, product discovery, architecture, and safe publishing
• Works with Claude Code and Codex; MIT licensed

• glebis/claude-skills
• ~100 skills covering meeting pipelines, research, image generation, and publishing
• Also personal analytics and Claude Code operations
• Each skill is a self-contained SKILL.md folder; installable via plugin marketplace or npx skills add

• getsentry/skills
• Agent skills the Sentry team uses internally, in the open Agent Skills format
• Includes AGENTS.md/CLAUDE.md authoring, blog writing standards, and brand guidelines
• Install as a Claude Code plugin marketplace or via npx skills add getsentry/skills
• For Sentry setup and production debugging skills, see getsentry/sentry-for-ai

Development & Engineering

Skills focused on software development, code quality, and engineering workflows.

Core Development Skills

• musoyangrigor/gitx-skill
• Turns messy AI-generated changes into clean, reviewable Git history
• Splits mixed working-tree changes into logical Conventional Commits
• Safe branches, checks, pull/push, PRs, issues, secret scanning, and conflict resolution
• Works with Claude Code, OpenAI Codex, Cursor, and Agent Skills-compatible tools

• gyujeongion/claude-code-rootcause
• Turns "never do X" bans into positive process gates
• Audits your instruction file for ban bloat (rethink + deusex)

• gyujeongion/claude-code-sandbox-gate
• Verification gate before a change hits main
• Stops the agent shipping "it works now" patches unchecked

• gyujeongion/claude-code-request-modes
• Classifies each request and activates only the rules that apply
• Stops one global rulebook bleeding into everything

• gyujeongion/claude-code-cold-tester
• Runs a doc through a zero-context model
• Finds where a real newcomer would get stuck

• gyujeongion/claude-code-mix-compare
• git diff for audio mixes — time-align two versions, phase-invert the change
• Per-stem, per-band, stereo and phase, translated into engineer notes

• obra/superpowers
• 20+ battle-tested development skills
• TDD (Test-Driven Development) workflows
• Debugging patterns and collaboration tools
• Git workflows and best practices

• mattpocock/skills
• 17 professional dev workflow skills
• PRD (Product Requirements Document) writing
• Codebase architecture guidance
• Git guardrails and issue triage

• YoungApple/growth-retrospective-skill
• Personal growth retrospectives ranked by decision-velocity signals from git + chat history
• Tracks 5 categories: domain knowledge, human skills, work habits, meta-cognition, productivity rhythm
• Graduation markers required on every Tier 1/2 item so items leave the list instead of piling up
• Step 0 Action Audit blocks new scans if prior level-up actions are untouched (anti-busywork forcing function)
• Cross-agent: same SKILL.md works in Claude Code, OpenAI Codex CLI, and Google Antigravity
• Try in 5 seconds: git clone … && bash examples/demo-fixture/demo.sh (no LLM needed)

• ggrigo/align
• Personal evals trio for Claude Code and Cowork — capture, diagnose, synthesize
• /align rates each LLM claim in a local HTML form (6-shape canonical: correct / wrong / almost / needs-nuance / can't-verify / skipped); archives corrections as structured markdown
• /diagnose traces wrong claims back to the stale instruction that caused them — quoted-evidence citations + 0–100 confidence + threshold-80 filtering (read-only)
• /retro clusters patterns across sessions and opens patch PRs with per-patch human gates
• MIT, runs locally, no service. Single-plugin marketplace at ggrigo/align; installs via /plugin marketplace add ggrigo/align

• affaan-m/ECC
• Agent harness performance optimization
• Research-first development approach
• Memory management and security patterns

• SawyerHood/dev-browser
• Give Claude agents web browser capabilities
• Headless browser automation with Playwright
• Sandboxed QuickJS WASM environment
• CLI tool with full LLM usage guide
• Works with Claude Code, Amp, and Codex

• arun-mosai/claude-code-slice-skills
• Three composable skills (/slice, /slice-resume, /slice-build) for vertical-slice feature development
• Turns a one-line feature idea into 11 validated design docs (persona, journey, UI, API, data, security, observability, AI, seed, tests, index)
• Then implements the slice as migrations, handlers, UI, events, and tests
• 12 validator sub-agents enforce specificity; atomic cycle commit; project-stage-calibrated ceremony
• Stack-neutral structure with TypeScript/Postgres/Zod/React defaults and a porting guide for Django/Rails/Go/Next.js

• mturac/recsys-pipeline-architect
• Designs composable recommendation, ranking, and feed pipelines using the six-stage Source → Hydrator → Filter → Scorer → Selector → SideEffect framework
• Pattern popularized by xAI's open-sourced X For You algorithm — independent MIT reimplementation
• Walks Claude through eight clarifying steps and surfaces architectural trade-offs (multi-action vs single-score, candidate isolation vs joint, online vs offline batch)
• Emits runnable scaffolds for Strapi v5 (TypeScript / Jest), Go (with generics), or Python / FastAPI — all three example suites green
• Vendor-agnostic: works with Claude Code, Codex CLI, Cursor, Gemini CLI, and 13 more agents via npx skills add mturac/recsys-pipeline-architect

• addyosmani/agent-skills
• Production-grade engineering skills aimed at AI coding agents
• Targets Claude Code, Cursor, and Antigravity IDE workflows
• Shell-driven and MIT licensed
• Maintained by Addy Osmani

• EveryInc/compound-engineering-plugin
• Official Compound Engineering plugin for Claude Code, Codex, Cursor, and more
• Loop of /ce-brainstorm, /ce-plan, /ce-work, /ce-code-review, and /ce-compound
• /ce-strategy anchors a STRATEGY.md and /ce-product-pulse writes time-windowed pulse reports
• Published as @every-env/compound-plugin on npm; documented by Every Inc

• OthmanAdi/planning-with-files
• Claude Code skill implementing Manus-style persistent markdown planning
• Plans live as files so they survive context resets and span sessions
• Compatible with Claude Code, Hermes Agent, OpenClaw, Kilocode, Copilot, and Mastra
• Python-backed and MIT licensed

• bo-cao/breaking-coding-chaos
• Human-in-the-loop dual-loop control-plane skill suite for coding agents
• throughline (progress cockpit) → plan-spar (align+lock PLAN) → clean-cut (minimal implement + verify)
• Standard Agent Skills layout; works with Claude Code, Codex, Cursor, Grok, OpenCode, Hermes, OpenClaw
• MIT licensed

• mturac/pluginpool
• Ten focused Claude Code plugins for everyday developer productivity
• Includes commit-narrator, pr-storyteller, test-gap, deps-doctor, env-lint, secret-guard, standup-gen, todo-harvest, flaky-detector, changelog-forge
• 89 hermetic tests across the suite, Python standard library only at runtime
• Each plugin is a self-contained repo; install all via marketplace: /plugin marketplace add mturac/pluginpool. MIT licensed.

• innovestrum/keelson
• Tracker-agnostic, issue-driven operating model for AI coding agents
• adopt-keelson interviews you and writes an AGENTS.md plus a selectable lifecycle skill-pack
• tune-gates establishes or refines just the quality gates, standalone
• Agents run mechanical work autonomously but escalate any move touching the original design, plan, or strategy to a human
• Works with GitHub, GitLab, Jira, Linear, or any API/CLI/MCP. MIT licensed.

• laki-arnsteinlarsen/strategistkit-skills
• Socratic Code Reviewer — interrogates intent, edge cases, and trade-offs; a senior reviewer's questions, not a checklist
• Systematic Bug Debugger — root-cause debugging from symptoms and stack traces; handles heisenbugs and "works locally, fails in prod"
• TDD Loop Master — drives red-green-refactor: builds a test list, writes tests first, keeps the discipline tight
• Free and MIT-licensed; a taste of the larger StrategistKit catalog

• zenlee123/routerbase-agent-skills
• Three agent skills for using routerbase as an OpenAI-compatible API gateway
• Covers API integration, model routing and fallback planning, and media generation workflows
• SKILL.md-based, MIT licensed, and compatible with Claude Code, Codex, OpenClaw, and other agent skill consumers

• github/spec-kit
• GitHub's toolkit for Spec-Driven Development (spec → plan → tasks → implement)
• Drives agents from a written specification rather than ad-hoc prompts
• Works with Claude Code, Copilot, Cursor, and Gemini CLI; Python, MIT licensed

• eyaltoledano/claude-task-master
• AI task-management system that parses a PRD into structured, dependency-aware tasks
• Ships both a CLI and an MCP server; drives step-by-step implementation
• Integrates with Claude Code, Cursor, Windsurf, and Roo

• yamadashy/repomix
• Packs an entire repository into a single AI-friendly file
• Feed whole codebases to Claude and other LLMs for analysis
• Token counting, compression, and remote-repo support; TypeScript, MIT licensed

• Pimzino/claude-code-spec-workflow
• Spec-driven workflow: Requirements → Design → Tasks → Implementation
• Plus a Report → Analyze → Fix → Verify bug workflow, with a live dashboard
• TypeScript, MIT licensed

• awesome-skills/code-review-skill
• Comprehensive single code-review Skill for Claude Code
• Covers 20+ languages/frameworks (React 19, Vue 3, Rust, TypeScript, TanStack Query v5)
• ~21k lines of review guidelines; MIT licensed

• carlrannaberg/claudekit
• Toolkit of custom slash commands, hooks, and utilities for Claude Code
• Hooks enforce type-checking and lint on edits
• TypeScript, MIT licensed

Specialized Agents

• VoltAgent/awesome-claude-code-subagents
• 100+ specialized subagents
• Full-stack development coverage
• Research and DevOps automation

• vijaythecoder/awesome-claude-agents
• 24 specialized development agents
• Tech Lead Orchestrator for team coordination
• Code Archaeologist for legacy code analysis
• Performance Optimizer for bottleneck detection

• lst97/claude-code-sub-agents
• Domain-expert subagents
• Multi-agent orchestration capabilities
• Automatic task delegation

• Jeffallan/claude-skills
• 66 specialized skills for full-stack developers
• Turns Claude Code into a domain-aware pair programmer
• Python-driven; covers frontend, backend, data, and ops

• Orchestra-Research/AI-Research-SKILLs
• Open-source library of AI research and engineering skills
• Vendor-neutral: works with Claude Code, Codex, and Gemini CLI
• LaTeX-driven docs; built for research-oriented teams

• revfactory/harness
• Meta-skill that designs domain-specific agent teams
• Generates specialized agents and the skills they use
• HTML-based interface for team composition

• 0xfurai/claude-code-subagents
• 100+ production-ready development subagents
• Roughly one per language, framework, or domain
• Drop into ~/.claude/agents/; MIT licensed

• hesreallyhim/a-list-of-claude-code-agents
• Community-submitted directory of Claude Code subagents
• Companion to the maintainer's awesome-claude-code list
• Browse and contribute agents by category

• qdhenry/Claude-Command-Suite
• Professional slash-command suite for structured dev workflows
• Code review, feature creation, security audit, and architecture commands
• Bundles agents and skills alongside the commands

• wshobson/commands
• Production-ready slash-command collection for Claude Code
• Split into multi-step workflows and single-purpose tools
• Companion to the popular wshobson/agents; MIT licensed

• composio-community/awesome-claude-plugins
• Curated list of Claude Code plugins and marketplaces
• Custom commands, agents, hooks, and MCP servers via the plugin system
• Community-maintained directory

Framework-Specific

• nextlevelbuilder/ui-ux-pro-max-skill
• Professional UI/UX design skill
• Modern design patterns and accessibility
• Component library integration

• conorluddy/ios-simulator-skill
• iOS Simulator automation for Claude Code
• 21 production scripts for building and testing
• Semantic navigation using accessibility APIs
• 96% token reduction vs raw tools
• CI/CD ready with JSON output

• IvanMurzak/Unity-MCP
• AI skills, MCP tools, and CLI for the Unity engine
• Full AI develop-and-test loop with efficient token usage
• C# implementation; quick CLI-based setup

• htdt/godogen
• Autonomous game development for Godot and Bevy
• Designed to run with Claude Code and Codex
• Python-driven; assists across engine, gameplay, and asset workflows

• coder/claudecode.nvim
• Claude Code Neovim IDE extension
• Embeds Claude Code into the Neovim workflow
• Lua-based; maintained by Coder

Code Review

• alibaba/open-code-review
• Hybrid code review: deterministic pipelines plus an LLM agent, with precise line-level comments
• Built-in multi-language ruleset covering NPE, thread-safety, XSS, and SQL injection
• OpenAI- and Anthropic-compatible; Go binary, also on npm as @alibaba-group/open-code-review
• Battle-tested at Alibaba scale

• The-PR-Agent/pr-agent
• The original open-source AI PR reviewer, now community-maintained
• Works across GitHub, GitLab, Bitbucket, and Azure DevOps
• Python-driven; a legacy Qodo project, distinct from Qodo's own offering

• coderabbitai/skills
• CodeRabbit's agent-native skills and plugin packaging for AI code review
• Installs into 35+ agents including Claude Code, Cursor, Gemini CLI, and Antigravity CLI
• Needs the CodeRabbit CLI; npx skills add coderabbitai/skills, MIT licensed

• modem-dev/hunk
• Review-first terminal diff viewer built for agent-authored changesets
• Multi-file review stream with sidebar nav and inline AI/agent annotations beside the code
• Split, unified, and auto layouts; watch mode for file and Git-backed reviews
• TypeScript, built on OpenTUI and Pierre diffs; works as a Git difftool, MIT licensed

AI Slop Elimination

Skills and tools that stop agent output from looking and reading machine-generated: design tells, generic copy, and the dead code, duplication, and complexity that agents leave behind. Writing-slop skills (blader/humanizer, op7418/Humanizer-zh) live in Marketing & Content.

Design & UI Slop

• Nutlope/hallmark
• Design skill for Claude Code, Cursor, and Codex that refuses the defaults every LLM was trained into
• Picks a macrostructure per brief, dresses it in one of 21 themes, runs 57 slop-test gates plus a pre-emit self-critique
• Four verbs: build (default), audit (score, no edits), redesign (keep copy and IA, new fingerprint), study (extract DNA from a design you admire)
• Made by Together AI; live demo at usehallmark.com

• miqdadbadjuber/anti-slop
• 38 mandatory rules (R-01 to R-38) in three tiers: Hard Gate, Purpose-Gate, Quality Locks
• A filter, not a style guide: no prescribed colors, fonts, or layouts — your DESIGN.md supplies direction
• Covers UI, copy, and code; ships a Liveliness Toolkit (ENERGY / RHYTHM / MOTION) so results stay specific instead of sterile
• Mandatory PASS/FAIL Delivery Gate before anything ships; additive skills so agents load only what a task needs

• funboy322/avoid-ai-design
• Audits AI-generated frontend, then rewrites it around a direction drawn from your product
• Catches both slop generations: purple-gradient hero era and the "tasteful" replacements (cream + terracotta, mono chrome, 01 / 02 / 03)
• Zero-dependency scanner included; writes a DESIGN.md so page two cannot drift from page one
• Design counterpart to avoid-ai-writing; MIT licensed, works with Claude Code, Cursor, Codex, and Copilot

• superdesigndev/superdesign-skill
• Design skill that generates and iterates UI, slide decks, and graphics on an infinite canvas
• Designs into your existing design system: reads your code for context and produces branchable drafts
• Cross-session continuity — remembers project, draft, and extracted components between runs
• Install with npx skills add superdesigndev/superdesign-skill, or as a Claude Code plugin; powered by superdesign.dev

• jakubkrehel/make-interfaces-feel-better
• Agent skill for the small details that separate shipped UI from generated UI
• Covers animations, typography, icons, hover states, optical alignment, concentric border radius, shadows, and hit areas
• Single SKILL.md; install with npx skills add jakubkrehel/make-interfaces-feel-better

Code Slop & Dead Code

• peteromallet/desloppify
• Agent harness that pairs mechanical detection (dead code, duplication, complexity) with LLM review of naming, abstractions, and module boundaries
• Prioritized fix loop with state that persists across scans, so quality improves over multiple sessions
• Scoring designed to resist gaming, plus a scorecard badge for your README
• 29 languages; installs a workflow guide for Claude, Cursor, Codex, Copilot, and others via desloppify update-skill

• millionco/react-doctor
• Deterministic scanner for the React your agent just wrote: state and effects, performance, architecture, security, a11y, maintainability
• Flags overly complex components and repeated JSX trees as composition candidates
• npx react-doctor@latest install turns the audit into a skill so the agent learns the fixes
• Works across Next.js, Vite, Astro, TanStack, React Native, and Expo; CI mode reports only newly introduced issues

• webpro-nl/knip
• Finds and fixes unused files, dependencies, and exports in JavaScript and TypeScript projects
• Catches the leftovers agents accumulate across refactors; ISC licensed
• Ships an MCP server (@knip/mcp), a language server, and VS Code and Open VSX extensions

• fallow-rs/fallow
• Single-binary static analysis for TypeScript and JavaScript: unused code, circular deps, duplication, complexity hotspots, boundary violations, design-system drift
• Deterministic findings and typed output contracts — no AI in the analyzer, no tsc or Node needed for static analysis
• Free static layer; optional paid Fallow Runtime adds production execution evidence

• kucherenko/jscpd
• Copy/paste detector for 224 formats, plus dead code, complexity hotspots, duplication trends over git history, and one health score
• Rust engine with per-language tokenization and Rabin-Karp hashing; opt-in Type-2 and Type-3 passes catch renamed and lightly edited copies
• AI-ready: MCP server, token-efficient reporter, SARIF/HTML/badge reporters, and a GitHub Action

Multi-Agent Systems

Orchestrate multiple Claude agents to work together on complex tasks.

Orchestration Platforms

• supernovae-st/nika
• Workflow engine for AI: Claude Code captures repeatable work as .nika.yaml DAG files
• Statically checked before execution (schema, permits, cost floor), tamper-evident traces after
• Ships a Claude Code plugin (authoring skill + read-only MCP oracle)

• gyujeongion/llm-council-no-api
• Claude chairs a panel of Gemini + GPT for a second opinion
• Runs through your logged-in CLIs, so no API keys

• gyujeongion/claude-code-goal-loop
• Autonomous goal loop with a verification gate each iteration
• Rollback snapshots and stall detection

• gastownhall/gastown
• Multi-agent workspace manager
• Persistent agent identity across sessions
• Convoy system for coordinated tasks
• Git-backed state tracking

• agentlas-ai/Agentlas-OS
• Open Agent OS for Claude Code, Codex, and Cursor
• Packages agents, skills, memory, routing rules, and security gates
• Includes Hephaestus Network MCP for Agentlas Hub search, runtime bundles, and plugin resolution
• Local-first by default with Hub fallback for public agents

• dlorenc/multiclaude
• Multi-agent orchestrator with flexible modes
• Singleplayer and multiplayer coordination
• Supervisor-subagent pattern implementation

• Untrivial-ai/agent-orchestrator
• Agent-agnostic parallel orchestrator
• Git worktrees for isolated environments
• CI auto-fix and review routing

• nyldn/claude-octopus
• Surface AI blindspots by running up to 8 AI models on every task
• Targets research, design, and coding workflows
• Shell-based orchestration layer

• RunMaestro/Maestro
• Agent orchestration command center
• Coordinates multi-agent runs across coding tasks
• TypeScript-driven; designed for fleet-scale agent operations

• stellarlinkco/myclaude
• Multi-agent orchestration workflow across Claude Code, Codex, Gemini, and OpenCode
• Go-based runtime distributed as a single binary
• Targets cross-agent collaboration on the same task

• generalaction/emdash
• Open-source agentic development environment (Y Combinator W26)
• Runs multiple coding agents in parallel across any provider
• TypeScript-driven

• ymxlx/polis-protocol
• Markdown coordination protocol for multi-vendor agent teams
• Signed capability cards, ε-greedy bandit routing that learns from settled contracts
• Chavruta paired review for high-stakes work; self-amending constitution
• Vendor-agnostic — works across Claude Code, Codex, Gemini CLI

• Varalix-Digitech-Solutions/clone-team
• Agent team (Manager, Frontend Developer, Backend Architect, Tester) that clones any website into a pixel-perfect UI in your chosen stack
• Deterministic build/test Workflow with an unskippable Tester gate; agent fan-out sized to host RAM
• Pausable/resumable across sessions; also outputs a reverse-engineered ARCHITECTURE.md

• Varalix-Digitech-Solutions/game-build-team-skill
• Agent team (Manager, Creative Director, Logic & Animation Developers, Tester) that builds Godot 4 / GDScript game features
• Two unskippable gates — headless test gate + "is it fun" creative gate — then an on-device final pass
• Deterministic build/test Workflow; pausable/resumable across sessions and usage-limit cutoffs

• 5dive-ai/5dive
• Run a company of AI coding agents on a server you own
• One-command spin-up of named agents (Claude Code, Codex, Grok, and more)
• Cron + heartbeat scheduling, shared task queue with human-escalation gates, Telegram control
• Babysit and needs-you triage dashboard; self-hosted, MIT

• ofekron/better-agent
• Local workspace for Claude, Codex, and Gemini coding-agent sessions
• Fork and delegate work across providers with persistent session state
• Detached runners and restart recovery keep long-running work inspectable
• Source-available for non-commercial use; commercial use requires permission

• beastlabai/multi-llm-plugin
• Multi-LLM orchestration plugin for Claude Code
• Parallel plan review, task generation, implementation, and code review across 12 coding harnesses (Codex, Cursor, opencode, pi, aider, Antigravity…)

• bmad-code-org/BMAD-METHOD
• "Breakthrough Method for Agile AI-Driven Development"
• Role-based agent team (analyst, PM, architect, dev, QA) in a plan-then-build flow
• Tool-agnostic framework used heavily with Claude Code, Cursor, and others

• BloopAI/vibe-kanban
• Kanban-style board to run and manage multiple coding agents
• Orchestrates Claude Code, Codex, and other agents from one UI
• Rust-based; Apache-2.0 licensed

• humanlayer/humanlayer
• Framework for steering coding agents through complex codebases
• Human-in-the-loop approvals and advanced context-engineering workflows
• TypeScript-based

• stravu/crystal
• Desktop app to run multiple Claude Code / Codex sessions in parallel git worktrees
• Compare competing approaches side by side
• TypeScript, MIT licensed

Parallel Processing

• Dicklesworthstone/claude_code_agent_farm
• Run 20-50 Claude agents in parallel
• Lock-based coordination system
• tmux monitoring dashboard

• disler/infinite-agentic-loop
• Two-prompt system for spawning parallel agents
• Evolving iterations for complex problem-solving
• Self-improving agent patterns

• AutomateLab-tech/agency-os
• Run your work like an AI agency from a single Notion board
• Plans ideas into dependency-sequenced subtasks
• Dispatches each task to the right model (Haiku/Sonnet/Opus)
• Collects results back in Notion via MCP
• Pluggable to Claude Code, Cursor, Cline, or any MCP-compatible agent

• ruvnet/ruflo
• Multi-agent swarm orchestration
• Self-learning capabilities
• RAG (Retrieval-Augmented Generation) integration
• Hive mind topology for distributed intelligence

• Yeachan-Heo/oh-my-claudecode
• Teams-first multi-agent orchestration
• Zero configuration with intelligent defaults
• Natural language interface, no commands to memorize
• Automatic parallelization and cost optimization
• Real-time HUD statusline for visibility

• mikeyobrien/ralph-orchestrator
• Improved implementation of the Ralph Wiggum technique for autonomous AI agents
• Rust-based runtime for managing long-running agent loops
• Built for fleet-style autonomous coding

• michaelshimeles/ralphy
• Bash-style Ralph Wiggum setup that runs agents in a loop
• Supports Claude Code, Codex, OpenCode, Cursor agent, Qwen, and Droid
• TypeScript implementation; lightweight harness

• subsy/ralph-tui
• Terminal UI on top of the Ralph autonomous-loop technique
• TypeScript-driven; provides visibility into iteration cycles

• smtg-ai/claude-squad
• Manage multiple AI terminal agents like Claude Code, Codex, OpenCode, and Amp
• Go-based runner with per-agent session control
• Built for keeping several agents alive in parallel

• SeemSeam/claude_codex_bridge
• Visible multi-agent CLI teams for Claude, Codex, Gemini, OpenCode, and Droid
• Project memory plus tmux-based supervision
• Python-driven runtime

• avelikiy/great_cto
• Multi-agent SDLC plugin for Claude Code
• 34 specialist reviewer agents across 25 archetypes (fintech, voice-AI, clinical, robotics, and more)
• 10 compliance packs auto-attach by archetype (PCI, HIPAA, FDA SaMD, NYC AEDT, GDPR, etc.)
• Two human gates per feature plus memory feedback loop that persists decisions across sessions

• dazuiba/handoff
• Skill/subagent that delegates tasks to DeepSeek V4, Codex, or Opus without leaving your session
• Runs tasks in the background; result returns automatically when done
• Backed by handoff-cli; supports parallel tasks and session resume

• dagger/container-use
• Sandboxed, isolated development environments for coding agents
• Lets multiple agents work in parallel without touching the host
• Go-based; Apache-2.0 licensed

Security & Compliance

Professional-grade security skills for vulnerability detection, code auditing, and compliance.

Security Analysis

• GiulioDER/cca-audit
• 6-layer parallel code audit pipeline with non-overlapping scopes
• Runs 6 specialized LLM auditors (code quality, bugs, security, performance, docs, config)
• Deduplicates findings, auto-fixes P1+P2, re-verifies tests, architect review gate
• Three variants: Claude Code, Codex CLI, OpenRouter Python CLI

• moltenbit/should-i-care
• Single-CVE applicability triage: does this CVE actually affect your environment
• Reasons about exploitation conditions against a self-maintained environment profile, every verdict cited to its source
• Markdown-driven; built and tested on Claude Code and Codex

• wrsmith108/varlock-claude-skill
• Secure environment variable management
• Ensures secrets never appear in sessions, terminals, logs, or git
• Production-ready secret handling

• trailofbits/skills
• Trail of Bits Claude Code skills for security research and audit work
• Vulnerability detection patterns from real engagement experience
• Python-driven

• mukul975/Anthropic-Cybersecurity-Skills
• 754 structured cybersecurity skills for AI agents
• Mapped to MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND, and NIST AI RMF
• Python-driven; large-scale skill bundle

• vercel-labs/deepsec
• Security harness for finding vulnerabilities using coding agents
• TypeScript-based; integrates with the agent loop
• Maintained by Vercel Labs

• SimoneAvogadro/android-reverse-engineering-skill
• Claude Code skill for Android application reverse engineering
• Shell-driven tooling around standard RE workflows
• Useful for security researchers and CTF teams

• gadievron/raptor
• Turns Claude Code into a general-purpose offensive and defensive security agent
• Uses CLAUDE.md, sub-agents, and rules for opinionated workflows
• Python-driven

• Darkmoon - Open source (GPL-3.0) autonomous AI penetration testing platform covering web, API, Active Directory and Kubernetes, with proof of exploitation.
• LoRexxar/Kunlun-M
• Fully open-source static white-box scanner for PHP and JavaScript
• Semantic scanning combined with AI Agent integration (OpenClaw and Claude Code)
• Python-driven; useful for application-security teams

• SpicyNoodles3/claude-code-file-guard
• PreToolUse hook that blocks the agent from writing to files you declare protected
• Covers Edit/Write/NotebookEdit, any tool carrying a file_path, plus a Bash write-heuristic and a destructive-git blocklist
• One auditable bash script, jq-only, fail-closed; README is explicit about what it cannot stop
• Tested on ubuntu and macos (39-case harness + shellcheck)

• usestrix/strix
• Open-source AI pentesting tool: autonomous agents that find and validate app vulnerabilities
• Python-driven, Apache-licensed, with hosted Strix Cloud and enterprise tiers
• For authorized testing of systems you own or have permission to assess

• snyk/agent-scan
• Discovers and scans the agent components on your machine — agents, MCP servers, and skills
• Checks them for prompt injections and vulnerabilities
• Install via uvx or a standalone binary (no npm package); CLI output is explicitly experimental

• gitleaks/gitleaks
• Detects secrets in code and git history — catches keys an agent pasted into a commit
• Go binary with a GitHub Action and Docker image
• Feature complete: security patches only, with development moving to Betterleaks

Compliance & Auditing

• BehiSecc/awesome-claude-skills
• Security-focused skill collection
• Document handling with security in mind
• Development and data analysis tools

Marketing & Content

Skills for marketing professionals, content creators, and growth teams.

• humanpen/humanpen-skill
• Document-level AI humanizer for .docx/.pptx — whole file, selected passages, or detector-flagged text
• Keeps every fact, number, table and all formatting intact
• The document never enters the model's context — upload, then get a result path back
• Works with Claude Code, Codex, Cursor, and other SKILL.md clients

• cognyai/claude-code-marketing-skills
• 5 free marketing skills, no account required
• SEO Audit, Landing Page Review, Competitor Analysis
• Ad Copy Writer and Lead Qualification
• Works with Claude Code, Cursor, and Windsurf

• coreyhaines31/marketingskills
• 23 professional marketing skills
• CRO (Conversion Rate Optimization)
• Copywriting and SEO strategies
• Paid advertising and growth tactics

• nowork-studio/NotFair
• Open-source Claude Code skills for SEO, GEO, Google Ads, and Meta Ads (MIT)
• SEO skills: site analysis, keyword research, meta tags, schema markup, GEO optimization, and content writing
• Google Ads skills: audits, wasted-spend detection, search-term cleanup, keyword and bid management
• Meta Ads skills: ROAS analysis, creative fatigue, and audience overlap detection
• Connects to live data via Google Ads MCP, Meta Ads MCP, Google Search Console MCP, and Google Analytics (GA4) MCP

• salespeak-ai/buyer-eval-skill
• Structured B2B software vendor evaluation skill (MIT)
• Domain-expert questions per software category
• Engages vendor AI agents for verified