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
- 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.
-
- Official Claude Code agentic coding tool
- Lives in your terminal and understands your codebase
- AI-powered development assistant
- Built by Anthropic
-
- Official repository for Claude Agent Skills
- Includes document processing (docx, pdf, pptx, xlsx)
- Art and design capabilities
- MCP (Model Context Protocol) builder
-
- 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
-
- 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
-
- 192+ professional skills and agent plugins
- Engineering, marketing, product management
- Compliance and C-level advisory capabilities
- Compatible with 11 different tools
-
- Chinese directory for Claude Code Skills, agents, and plugins
- 140+ curated resources for Chinese-speaking developers
- Useful for discovering practical skills by category
-
- 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
-
- 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
-
- 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
-
- 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/antigravity-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
-
- 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-published skills, MCP servers, and Custom Agents
- Ships
AGENTS.mdbundles for grounding coding agents - TypeScript-driven; useful starting point for enterprise teams
-
- Anthony Fu's curated collection of agent skills
- Skills cover web tooling, build configuration, and developer ergonomics
- TypeScript implementation
-
- 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
-
- 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
-
- Hub for Claude Skills, Agents, Commands, Hooks, Plugins, and Marketplace collections
- Extends Claude Code and Claude Desktop
- Python-driven discovery layer
-
- 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.mdconfigs - Templates organized into 19 categories
- HTML-driven gallery
- 162 production-ready AI agent templates with
-
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
-
- 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
-
- 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)
Development & Engineering
Skills focused on software development, code quality, and engineering workflows.
Core Development Skills
-
- 20+ battle-tested development skills
- TDD (Test-Driven Development) workflows
- Debugging patterns and collaboration tools
- Git workflows and best practices
-
- 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)
-
- Personal evals trio for Claude Code and Cowork — capture, diagnose, synthesize
/alignrates 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/diagnosetraces wrong claims back to the stale instruction that caused them — quoted-evidence citations + 0–100 confidence + threshold-80 filtering (read-only)/retroclusters 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/everything-claude-code
- Agent harness performance optimization
- Research-first development approach
- Memory management and security patterns
-
- 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
- Three composable skills (
-
- Post-generation quality analyzer for AI-written code — 27 rules across 7 dimensions (dead code, duplication, architecture, error handling, dependencies, security, testing)
- Targets AI-native smells that ESLint/SonarQube miss: over-generation leftovers, structural clones, empty except blocks, god files, unused deps
sober scan .prints a 0–100 sobriety score with severity-bucketed findings;sober fix <rule-id>returns actionable fix instructions per rule- Language-agnostic and zero-config; CI mode returns non-zero on threshold breaches
- Ships Claude Code slash commands so reviews run as a /command inside the editor
-
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
-
- 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-strategyanchors aSTRATEGY.mdand/ce-product-pulsewrites time-windowed pulse reports- Published as
@every-env/compound-pluginon npm; documented by Every Inc
-
- 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
-
- 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.
-
- Tracker-agnostic, issue-driven operating model for AI coding agents
adopt-keelsoninterviews you and writes an AGENTS.md plus a selectable lifecycle skill-packtune-gatesestablishes 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
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
-
- Domain-expert subagents
- Multi-agent orchestration capabilities
- Automatic task delegation
-
- 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
-
- Meta-skill that designs domain-specific agent teams
- Generates specialized agents and the skills they use
- HTML-based interface for team composition
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
-
- 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
-
- Autonomous game development for Godot and Bevy
- Designed to run with Claude Code and Codex
- Python-driven; assists across engine, gameplay, and asset workflows
-
- Claude Code Neovim IDE extension
- Embeds Claude Code into the Neovim workflow
- Lua-based; maintained by Coder
-
infai-tech/claudemd-auditor-skill
- Calibrated 0–100 audit of
CLAUDE.mdfiles on six axes (Specificity, Coverage, Brevity, Currency, Quirks captured, Tone fit) - v1.0 matches human baselines at Δ=0 across a 6-case corpus spanning 14–78 points; adversarial corpus exercises three error codes (
too_short,not_a_claudemd,too_long) - Returns paste-ready Top-3 improvements with concrete markdown snippets
- Drop-in
~/.claude/skills/claudemd-auditor/install; MIT licensed; no external dependencies
- Calibrated 0–100 audit of
Multi-Agent Systems
Orchestrate multiple Claude agents to work together on complex tasks.
Orchestration Platforms
-
- Multi-agent workspace manager
- Persistent agent identity across sessions
- Convoy system for coordinated tasks
- Git-backed state tracking
-
- 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
-
- Multi-agent orchestrator with flexible modes
- Singleplayer and multiplayer coordination
- Supervisor-subagent pattern implementation
-
- Agent-agnostic parallel orchestrator
- Git worktrees for isolated environments
- CI auto-fix and review routing
-
- Surface AI blindspots by running up to 8 AI models on every task
- Targets research, design, and coding workflows
- Shell-based orchestration layer
-
- Agent orchestration command center
- Coordinates multi-agent runs across coding tasks
- TypeScript-driven; designed for fleet-scale agent operations
-
- 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
-
- Open-source agentic development environment (Y Combinator W26)
- Runs multiple coding agents in parallel across any provider
- TypeScript-driven
-
- Orchestration layer for coding agents including Claude Code and Codex
- One prompt can drive multiple downstream agents
- TypeScript-driven
-
yehudalevy-collab/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
Parallel Processing
-
Dicklesworthstone/claude_code_agent_farm
- Run 20-50 Claude agents in parallel
- Lock-based coordination system
- tmux monitoring dashboard
-
- Two-prompt system for spawning parallel agents
- Evolving iterations for complex problem-solving
- Self-improving agent patterns
-
- 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
-
- Multi-agent swarm orchestration
- Self-learning capabilities
- RAG (Retrieval-Augmented Generation) integration
- Hive mind topology for distributed intelligence
-
- 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
-
- Real-time multi-AI collaboration bridge
- Claude, Codex, Gemini collaboration with persistent context
- Split-pane terminal with WYSIWYG interface
- Minimal token overhead and async messaging
- Every interaction visible and controllable
-
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
-
- 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
-
- Terminal UI on top of the Ralph autonomous-loop technique
- TypeScript-driven; provides visibility into iteration cycles
-
- 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
-
- Visible multi-agent CLI teams for Claude, Codex, Gemini, OpenCode, and Droid
- Project memory plus tmux-based supervision
- Python-driven runtime
-
- 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
Security & Compliance
Professional-grade security skills for vulnerability detection, code auditing, and compliance.
Security Analysis
-
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
-
- 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
-
- 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
-
- 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
-
- Turns Claude Code into a general-purpose offensive and defensive security agent
- Uses
CLAUDE.md, sub-agents, and rules for opinionated workflows - Python-driven
-
- 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
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.
-
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
-
- 23 professional marketing skills
- CRO (Conversion Rate Optimization)
- Copywriting and SEO strategies
- Paid advertising and growth tactics
-
- Open-source Claude Code plugin for SEO, SEM, and Google Ads
- Connects Search Console and Google Ads for actionable audits
- Includes SEO analysis, content writing, and CMS setup skills
-
- 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
-
- Structured B2B software vendor evaluation skill (MIT)
- Domain-expert questions per software category
- Engages vendor AI agents for verified due diligence
- Evidence-tracked scoring across 7 dimensions; comparative scorecards for procurement and build-vs-buy decisions
-
zarazhangrui/codebase-to-course
- Transform codebases into interactive HTML courses
- Beautiful, educational content generation
- Perfect for developer education and onboarding
-
- Schedule and publish social media posts across 13 platforms via SocialClaw
- Supports X, LinkedIn (profile + page), Instagram, Facebook Pages, TikTok, YouTube, Reddit, WordPress, Pinterest
- One workspace API key (
SC_API_KEY) for campaigns, media upload, and analytics - Install:
npx skills add ndesv21/socialclaw
-
- Agent skill for operating Sequenzy lifecycle email marketing and transactional/product email workflows
- Manage subscribers, lists, segments, tags, templates, campaigns, sequences, stats, API keys, websites, and transactional sends
- Includes CLI/MCP decision trees, supported-workflow caveats, dashboard URL patterns, and safe execution guidance for AI agents
- Install:
npx skills add Sequenzy/skills
-
- Open-source SKILL.md pack for newsletter operators using Claude Code, Codex, Cursor, and compatible agents
- Covers local event research, source audits, issue building, sponsor operations, monetization strategy, ROI checks, social repurposing, and launch cadence
- Product-agnostic by default, with optional Maito-connected persistence for sources, issues, approvals, analytics, and sponsor history
- Instruction-only v0.1.0 pack with no executable scripts or runtime dependencies
- Install:
npx skills add Get-Maito/newsletter-skills
-
- Turn any file, folder, URL, or service export into a polished single-file HTML page
- Auto picks one of 16 design systems (lesson lab, map atlas, timeline story, relationship rhythm, network map, dashboard, document review, terminal evidence, living essay, editorial carousel, paper trail, …)
- 34 source parsers — CSV, PDF, DOCX, JSON/JSONL, log, GPX/KML, WhatsApp/WeChat/Slack/Discord/Telegram, Amazon orders, Kindle highlights, Spotify, Apple Health, LinkedIn, Google Photos Takeout, vCard, Venmo/PayPal, browser history, …
- Self-contained — inline CSS/JS, no CDN, works offline; 22-demo live gallery
- Cross-agent — Claude Code, Codex, Cursor, Cline, Windsurf via
npx skills add clockless-org/html-anything
-
- Paid advertising audit and optimization skill for Claude Code
- 250+ checks across Google, Meta, YouTube, LinkedIn, and TikTok ads
- Python-driven
-
- Universal SEO skill for Claude Code
- 25 sub-skills plus 18 sub-agents covering technical SEO, E-E-A-T, schema, GEO/AEO, backlinks, and local SEO
- Python-driven
-
zubair-trabzada/geo-seo-claude
- GEO-first SEO skill for Claude Code
- Citability scoring, AI crawler analysis, and brand-mention tracking
- Python-driven
-
- Agent skill for generating polished HTML slide decks
- Editorial-magazine and Swiss layouts with social-cover variants
- Image prompts and WebGL/low-poly visual options
-
- Local-first, open-source alternative to Anthropic's Claude Design
- 19 skills with 71 brand-grade design systems
- Generates web, desktop, and mobile prototypes; BYOK
-
- HTML-native design skill for Claude Code
- High-fidelity prototypes, slide decks, and animation with MP4 export
- 20 design philosophies and a 5-axis review; agent-agnostic
-
- Open-source AI-native vector design tool
- First to feature concurrent agent teams for design-as-code
- TypeScript-driven
-
yizhiyanhua-ai/fireworks-tech-graph
- Generate production-quality SVG and PNG technical diagrams from natural language
- 7 styles, UML support, and AI/Agent workflow patterns
- Python-driven
-
aislinn-yang/rednote-writing-skill
- Writes Rednote (Xiaohongshu / 小红书) posts in your own voice, never fabricating content
- Collage / co-create / review modes turn rough ideas into a finished draft
- Scans for AI-tells and platform-throttling words, then a "reads like a real person" pass before title and cover
- MIT licensed, instruction-only (no executable scripts); tune
examples.mdto match your own writing
-
- GTM pipeline agent that syncs Smartlead, Instantly, HeyReach, PlusVibe, EmailBison, Prosp, and Calendly into a local SQLite database your agent can query directly
- Every send, reply, bounce, and stage change lands on your machine
- Install:
npx skills add outreachmagic/outreachmagic
Domain-Specific Skills
Specialized skills for specific industries and use cases.
-
Donchitos/Claude-Code-Game-Studios
- Turn Claude Code into a full game development studio
- 48 AI agents and 37 workflow skills
- Complete coordination system mirroring real studio hierarchy
- Engine specialists, designers, and production agents
- Perfect for game development projects
-
- Complete Minecraft Paper plugin development suite for Claude Code
- 1 comprehensive skill + 8 specialized subagents (architect, debugger, reviewer, data, commands, events, performance, publisher)
- Enforces modern practices: Adventure API, Display entities, config-driven items/GUIs, lang files, LuckPerms permission trees
- MythicMobs, ModelEngine, PlaceholderAPI, WorldGuard, Vault, LuckPerms integrations
- Kotlin/Java, Paper 1.21.x, Gradle Kotlin DSL
-
fruitwyatt/puzzlegenio-claude-skill
- Generate free printable puzzles (crossword, word search, sudoku, jigsaw, bingo, nonogram)
- Deep links with URL parameters for pre-filled puzzle settings
- 8 languages with native word generation
- AI-powered clue and word list generation
- Perfect for teachers, event planners, and makers
-
pedrohcgs/claude-code-my-workflow
- Ready-to-fork template for academic research
- LaTeX/Beamer + R workflow optimization
- Multi-agent review and adversarial QA
- Quality gates and replication protocols
- Extracted from production PhD course
-
- Ecommerce business review skill for D2C stores
- Multi-horizon KPI decomposition across 30d/90d/365d
- ~30 automated health checks across revenue, customer, and product
- RFM cohorts and prioritized action plan with impact estimates
- Hybrid Python-backend + Claude-LLM architecture for deterministic KPIs
-
- Business research skill for Claude Code and Codex
- Encodes 19 consulting frameworks and 9 thinking methods
- Uses evidence grading, source confidence checks, and stage gates to reduce unsupported AI research claims
- Generates decision-ready HTML reports with interactive charts
- MIT licensed; installable from GitHub
-
- OpenClaw plugin and agent skill for X/Twitter automation
- Search tweets and tweet replies, post tweets and replies, and export followers
- Includes media upload, media download, direct messages, monitors, webhooks, and giveaway draws
- Ships GitHub docs, npm install, ClawHub listing, and Context7 reference docs
-
- Native Hermes Agent plugin for X/Twitter research, monitoring, and gated actions
- Ships a PyPI package, Hermes plugin entry point, bundled skill, and
.claude-plugin/plugin.json - Install with
hermes plugins install Xquik-dev/hermes-tweet --enable - Read tools use
XQUIK_API_KEY; write actions also requireHERMES_TWEET_ENABLE_ACTIONS=true
-
- Detector + rewriter for Chinese AI-generated text — pure Python, zero dependencies
- 0–100 statistical scorer across 11 weighted dimensions (discourse density, parallel structure, four-character cliché, hedging, etc.) plus sentence-level breakdown
- 7 style transforms (casual / zhihu / xiaohongshu / wechat / academic / literary / weibo) with genre-aware auto-routing for long-form text
- Targeted optimizations for Chinese academic platforms (CNKI / VIP / Wanfang / 朱雀) with hedging injection and four-char cluster smoothing
- Ships as a Claude Code skill plus standalone CLI; offline-first, no API keys
-
- Claude Code and OpenCode skill that removes signs of AI-generated writing
- Installs by cloning into
~/.claude/skills/humanizeror~/.config/opencode/skills/humanizer - Single-file
SKILL.mdinvoked via/humanizerin either tool - MIT licensed
-
- End-to-end job-search system built on Claude Code with 14 skill modes
- Resume tailoring, batch outreach, interview prep, and PDF generation
- Go dashboard for tracking applications and pipeline status
- MIT licensed; site at career-ops.org
-
- Agent skill that researches topics across Reddit, X, YouTube, Hacker News, TikTok, Polymarket, Bluesky, and the open web
- Synthesizes findings into a grounded recency summary, weighted toward the last 30 days
- Targets deep research, social-media monitoring, and trend tracking
- Compatible with Claude Code, OpenClaw, and ClawHub
-
K-Dense-AI/scientific-agent-skills
- Ready-to-use agent skills for research, science, engineering, analysis, finance, and writing
- Coverage includes bioinformatics, chemoinformatics, genomics, proteomics, and materials science
- Targets Claude Code and Claude Skills runtimes
- Maintained by K-Dense AI; MIT licensed
-
FreedomIntelligence/OpenClaw-Medical-Skills
- Open-source medical AI skills library for OpenClaw
- Skills tuned for clinical, research, and pharma workflows
- Python-driven
-
Imbad0202/academic-research-skills
- Academic research skills for Claude Code
- Research, write, review, revise, finalize pipeline
- Python-driven
-
Master-cai/Research-Paper-Writing-Skills
- Skill package for ML, CV, and NLP paper writing
- Adapted from Prof. Peng Sida's open notes
- Works with Codex, Claude Code, and Gemini CLI
-
- Semi-automated research assistant for academic and software work
- Supports Claude Code, OpenCode, and Codex CLI
- Python-driven
-
SamurAIGPT/Generative-Media-Skills
- Multi-modal generative media skills for AI agents
- High-quality image, video, and audio generation
- Works with Claude Code, Cursor, and Gemini CLI
-
- Distill anyone's thinking — mental models, decision heuristics, and expression DNA
- Python-driven Claude skill
- English/Chinese bilingual
-
- Chinese localization of the Humanizer Claude Code skill
- Removes AI-generated signals from Chinese text
- Companion to blader/humanizer for English
-
- Skill that lets your skills evolve through evaluate, improve, test, and keep-or-rollback cycles
- Autoresearch-inspired iteration loop
- HTML-driven; Chinese-language project
-
- Claude skill that writes accurate prompts for any AI tool
- Aims to keep context and memory across iterations
- Token-conscious by design
-
- 125+ agent skills for Longbridge Securities — real-time quotes, charts, fundamentals, portfolio analytics, and valuation
- Covers HK, US, A-share (SH/SZ), and Singapore markets
- Trilingual support: Simplified Chinese, Traditional Chinese, and English
- Includes options (US/HK), warrants, DeFi yields, ETF analysis, and advanced portfolio tools
-
- A reasoning-lens skill that makes any AI advise through the Three-Pillar system (Income − Expenses = Freedom Fund; Cashflow, Investment, Savings; discipline over strategy)
- MIT licensed; installable via plugin/degit
-
- Solana-native trust scoring MCP server for AI agents
- Free preflight endpoint + paid signed V5 trust receipt via x402 protocol
- Streamable-HTTP MCP at https://intel.twzrd.xyz/mcp
- On-chain trust verification and payment receipts for Web3 workflows
-
- Bilingual EN/中文 skill that rewrites a rough idea or over-specified shot script into a model-ready text-to-video prompt
- Works with Seedance 2.0, Kling, Veo, Hailuo, Wan, Vidu
- Judgment-first: won't flatten prompts that need precise control
- One SKILL.md, open Agent Skills standard; MIT
Productivity Tools
Utilities and tools to enhance your Claude workflow.
Desktop Applications & GUI Tools
-
- AI productivity studio with smart chat and autonomous agents
- 300+ built-in assistants for various tasks
- Unified access to frontier LLMs
- Cross-platform desktop application
-
- All-in-one desktop assistant for Claude Code, Codex, OpenCode
- Skills management and provider configuration
- Cross-platform with WSL support
- Built with Rust and Tauri for performance
-
- Powerful GUI app and toolkit for Claude Code
- Create custom agents and manage interactive sessions
- Run secure background agents
- Built with Rust and Tauri
-
- Free, local, open-source 24/7 cowork app
- OpenClaw alternative for multiple AI coding assistants
- Supports Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code
- Web UI for easy access
-
- Desktop and mobile UI for Claude Code, Cursor CLI, Codex
- Manage sessions and projects remotely
- Responsive design works on desktop, tablet, and mobile
- Built-in chat, shell terminal, file explorer, and git integration
- Plugin system for custom extensions
-
- Native macOS app for searching, browsing, and resuming Claude Code sessions alongside Codex, Gemini CLI, OpenCode, and other agents
- Agent Cockpit provides live iTerm2 orchestration and monitoring for active local agent sessions
- Local-first transcript indexing with no telemetry
-
- Local, open-source macOS app for listening to Claude Code and Codex agents as they work
- Global Mix across active sessions, blocker alerts, and BYOK Gemini/OpenAI narration
-
- Native macOS notch companion for Claude Code and Codex long runs
- Shows live session state and usage locally
- Optional auto-resume for trusted sessions after reset windows
-
- Native macOS menu-bar app to use Claude Code and ChatGPT subscriptions with AI coding tools
- Skips the need for API keys
- Swift implementation
-
- Desktop app providing a graphical interface for OpenClaw AI agents
- Turns CLI-based AI orchestration into a desktop experience
- TypeScript-driven Electron build
Agent Harnesses & Meta-Tools
-
- Best-in-class agent harness (previously oh-my-opencode)
- TUI for orchestration and management
- Multi-provider support (OpenAI, Gemini, Claude)
- TypeScript-based with excellent developer experience
-
- Lightweight meta-prompting and context engineering system
- Spec-driven development approach
- Created by TÂCHES team
- Focus on productivity and getting things done
-
- Applies the same rules to every coding agent on your machine
- One rule set propagates to Claude Code, Cursor, Codex, and others
- TypeScript-driven
-
- Local-first Agent Skills profile manager for grouping
SKILL.mddirectories by scenario - Switches lean profiles such as core, research, frontend, and CTF with backup and restore support
- Supports Claude Code, Codex, OpenCode, and generic local skill roots with JSON automation output
- Local-first Agent Skills profile manager for grouping
-
- Claude Code framework focused on saving time on routine prompts
- Built for developers tired of writing "please act like a senior engineer" every session
- Python-driven
-
- AI-enabled pair programmer with broad model support
- Works with Claude, GPT, Grok, DeepSeek, Gemini, and 300+ more models
- Rust-based binary
-
- CLI for managing Git worktrees in parallel AI-agent workflows
- Rust-based binary
- Reduces friction when running multiple coding agents on the same repo
-
huangji6693-max/claude-code-dna
- Behavioral OS for Claude Code agents distilled from 11 deeply-read libraries (179 skills + 99 agents + Karpathy anti-patterns + memory research)
- 8 rule files codifying Karpathy's 4 laws plus 15 operating instincts (verification gate, TDD reflex, root cause before fix)
- Three-layer memory architecture inspired by mem0 v3 (ADD-only), langmem (3 types × 2 timings), and Microsoft GraphRAG (community-summary indexing)
- Portable Bash/Python tooling:
memory-health.sh,memory-search.sh(BM25-style local retrieval),skill-spec-audit.sh(agentskills.io compliance — 168 PASS / 23 WARN / 3 FAIL on 194 community skills), anddna-doctor.sh(single-command health check) - Works across Claude Code, Cursor, Codex CLI, Gemini CLI, Aider, and Continue.dev — rules are model-agnostic markdown
-
qingqingpi/loop-engineering-skill
- Design, evaluate, diagnose, and harden repeated or unattended AI-agent loops, organized around verifier fidelity (a loop is only as good as its verifier)
- GREEN/YELLOW/RED task-fit triage with two hard vetoes; three-layer architecture (scheduler, inner loop, commit gate) keeps irreversible actions outside the loop body
- Separates task loop-fit from as-specified-design safety (a GREEN-fit task can still have an UNSAFE design); ships a reproducible cross-model eval with pre-registered held-out and minimal-pair suites, a standalone runner, and an honest limitations writeup
Memory & Context Management
-
- Privacy-safe context receipts for Claude Code, Cursor, Codex, OpenClaw, and MCP/skill workflows
- Proves what context crossed agent boundaries without copying private prompts or tool outputs
- Includes an npm CLI plus copyable Context Receipts skill recipe and 60-second smoke checks
- Focuses on audit evidence, not persistent memory or agent orchestration
-
maxbaluev/accreted-intelligence
- Local-first MCP work memory for Claude Code, Codex, Cursor, and OpenCode
- Scored retrieval, commitment tracking, and outcome-based credit help coding agents reuse lessons across sessions
- Public installer, docs, MCP/plugin integration glue, and release artifacts; private engine binary
-
- Local-first, markdown-first
.akephalospassport for portable agent preferences, tool notes, rules, project context, and durable memories - Works across Claude Code, Codex, Cursor, Hermes, OpenClaw, MCP clients, and machines
- Sync is via plain files/Git; includes a local MCP stdio server
- Early
v0.1.0prerelease
- Local-first, markdown-first
-
- Automatic session capture and AI compression
- Long-term memory with RAG integration
- Uses Claude's agent-sdk for intelligent context injection
- SQLite and ChromaDB for embeddings storage
- Seamless integration with future Claude sessions
-
- Make entire codebase the context for Claude Code
- Semantic code search MCP plugin
- Vector database for efficient codebase storage
- Cost-effective for large codebases
- No multi-round discovery needed
-
- Knowledge-graph skill that ingests code, SQL schemas, R/shell scripts, docs, papers, images, and videos
- One graph holds application code, database schema, and infrastructure side by side
- Uses tree-sitter parsing and Leiden community detection for GraphRAG
- Works with Claude Code, Codex, OpenCode, Cursor, and Gemini CLI
-
- Builds a persistent, incremental knowledge graph of your codebase for Claude Code
- Claims roughly 6.8x fewer tokens on reviews and up to 49x on daily coding tasks
- Tree-sitter static analysis exposed via MCP for graph queries
- Python and MIT licensed; project site at code-review-graph.com
-
- Sandboxes tool output to compress context for AI coding agents
- Claims up to 98% context reduction across long sessions
- Ships hooks, skills, and an MCP server for 15+ platforms (Claude Code, Codex, Cursor, Kiro, Zed, OpenClaw)
- Project site at context-mode.com
-
- Memory upgrade for coding agents
- Persists context across sessions for Claude Code and other agents
- Go-based binary
-
- Memory layer for 24/7 proactive agents like OpenClaw
- Python-driven graph-based memory store
- Targets long-running agent workflows
-
- Personal knowledge base that builds and maintains itself
- Drop in sources and Claude (or Codex/Gemini) extracts and links knowledge
- Python-driven
-
- Context management for Claude Code via hooks
- Maintains state via ledgers and handoffs across sessions
- MCP execution without context pollution
-
- Obsidian vault that gives AI coding agents persistent memory
- Designed for Claude Code, Codex CLI, and Gemini CLI
- TypeScript-driven
-
- High-performance code-intelligence MCP server
- Indexes codebases into a persistent knowledge graph in milliseconds
- C implementation for speed
-
- Claude Code plugin that builds individualized knowledge systems from conversation
- You describe how you think and work; it captures and reuses that context
- Shell-driven
-
- Local proxy that compresses each model request to cut token cost, with quality gating that reverts any step that does not save
- Works with Claude Code via HTTPS_PROXY with no code changes
- Trims resent tool schemas, history, MCP tool output, and scraped pages on every turn
- Also ships an MCP server, CLI, libraries (Rust, Python, Ruby, Swift, Kotlin), and a WASM/JS npm package
- Rust-based, MPL-2.0 licensed
-
- Agent-agnostic context layer: a plain git repo of Markdown + YAML your coding agent reads at the start of every session
- It begins each session knowing your repos, clients, and task conventions instead of restarting from zero
- One
skills/tree works across Claude Code, Codex, and Copilot CLI; sync is via plain files/Git - MIT licensed; single self-hosted instance so far — documents the method, not adoption
MCP Servers & Integrations
-
- Powerful coding agent toolkit with semantic capabilities
- MCP server for retrieval and editing
- Language server protocol integration
- Advanced semantic code understanding
-
- Lightweight OpenClaw alternative for containers
- Connects to WhatsApp, Telegram, Slack, Discord, Gmail
- Built on Anthropic's Agents SDK
- Memory and scheduled jobs support
-
RichardAtCT/claude-code-telegram
- Telegram bot for remote Claude Code access
- Chat naturally about projects from anywhere
- Automatic session persistence per project
- Proactive notifications from webhooks and CI/CD
- Built-in authentication and audit logging
-
wonderwhy-er/DesktopCommanderMCP
- MCP server for terminal control and file system management
- Execute commands, search files, and diff-based editing
- Native Excel, PDF, and DOCX support
- Interactive process control and session management
- Docker isolation for sandboxing
-
- MCP server (
kc-agent) bridging AI assistants to multi-cluster Kubernetes - Manage workloads, pods, namespaces, and RBAC across clusters via natural language
- Built-in AI-powered dashboard with 250+ CNCF project integrations
- Works with Claude Code, Claude Desktop, and any MCP-compatible client
- Live demo: https://console.kubestellar.io
- MCP server (
-
- MCP integration that lets AI agents talk to Figma
- Supports Cursor and Claude Code for reading and editing designs
- JavaScript-based; published by Grab's engineering team
-
atilaahmettaner/tradingview-mcp
- MCP for real-time crypto and stock screening
- Advanced technical indicators and Bollinger-band intelligence
- Python implementation; works as a native Claude Code MCP
-
- MCP server for Google NotebookLM
- Lets agents (Claude Code, Codex) research documentation with grounded citations
- TypeScript-driven
-
google-labs-code/stitch-skills
- Library of agent skills designed to work with the Stitch MCP server
- Follows the open Agent Skills standard
- TypeScript-driven; published by Google Labs
-
- Bridges local AI coding agents to messaging platforms
- Supports Claude Code, Cursor, Gemini CLI, and Codex
- Targets Feishu/Lark, DingTalk, Slack, Telegram, and Discord
- Go-based runtime
-
- Bridges Claude Code and Codex to IM platforms
- Chat with AI coding agents from Telegram, Discord, or Feishu/Lark
- TypeScript implementation
-
- Adds full internet access to Claude Code via a layered skill
- Three-tier channel scheduling, browser CDP, and parallel divide-and-conquer
- JavaScript-driven
-
- Claude Code plugin (zero@zero-plugins) that gives your AI access to thousands of tools, APIs and services to use and pay for per use
- No config, no API keys; services are discovered and paid for at runtime
- Also ships a CLI (@zeroxyz/cli) and an MCP connector (mcp.zero.xyz)
- Homepage: https://zero.xyz
API & Integration Tools
-
- Wrap multiple AI CLIs as OpenAI-compatible API
- Support for Gemini CLI, Claude Code, ChatGPT Codex, Qwen Code
- Access free models through standardized API interface
- Built with Go for performance
-
- Run Claude Code on OpenAI and Gemini models
- Anthropic API proxy for multiple model providers
- Powered by LiteLLM for model routing
- Transparent proxy with custom model mapping
- Use any Anthropic client with alternative models
-
- Go reverse proxy exposing Anthropic, Gemini, and OpenAI-compatible APIs behind one local endpoint
- Multi-provider routing across GitHub Copilot, Azure OpenAI, and OpenAI Codex
- Zero-config Copilot mode plus JSON/YAML provider configs for explicit routing
- macOS/Linux menubar tray app and distroless container
- MIT licensed, single static Go binary, no frameworks
-
- Agent skill for generating AI images, videos, and music/audio from Claude Code and compatible SKILL.md agents
- Also covers JSON-first requests, async polling, and other model API jobs through the RunAPI CLI
- JSON-first requests, async task polling, and terminal/CI-friendly workflows
- Apache-2.0 licensed; install with
git clone https://github.com/runapi-ai/cli-skill ~/.claude/skills/runapi-cli
CLI Configuration Tools
-
- CLI tool for configuring and monitoring Claude Code
- Template management system
- Configuration automation
- Python-based for easy customization
-
- Use Claude Code as foundation for coding infrastructure
- Route requests to different models based on needs
- Multi-provider support (OpenRouter, DeepSeek, Ollama, Gemini)
- Dynamic model switching with /model command
- GitHub Actions integration
-
- CLI and MCP server that removes AI writing patterns from text
- Strips tricolons, em-dash overuse, sycophancy openers, and overused vocabulary
- Works with Claude Code, Codex, Gemini CLI, and Cursor via PostToolUse hooks
- Five intensity levels plus a lint-only audit mode
-
- Zero-Config Code Flow for Claude Code and Codex
- TypeScript-driven configuration wrapper
-
ainova-systems/intelligence-sync
- One source of truth for AI coding rules across every IDE
- Author rules, agents, and skills once in plain Markdown
- Routes them into each tool's native formats (Claude Code, Cursor, GitHub Copilot, Codex, AGENTS.md)
- Zero-dependency, bash and awk, MIT licensed
Autonomous Development
-
- Autonomous AI development loop with intelligent exit detection
- Continuous development cycles until project completion
- Built-in safeguards to prevent infinite loops
- Rate limiting and circuit breaker protection
- Global command available in any directory
-
- Build and evaluate agents with model, Skills, MCP, and ACP support
- Python-based research-grade harness
- Designed for fast agent-iteration cycles
Monitoring & Observability
-
disler/claude-code-hooks-multi-agent-observability
- Real-time agent monitoring
- Hook event tracking with SQLite dashboard
- Visual tool event indicators
-
Maciek-roboblog/Claude-Code-Usage-Monitor
- Real-time terminal monitoring for token usage
- Machine learning-based predictions and analytics
- Burn rate tracking and cost analysis
- Rich UI with intelligent session limit warnings
- Track consumption across different Claude plans
-
- macOS menu bar app for usage stats
- Track Codex, Claude, Cursor, Gemini, and more
- Session and weekly limits visibility
- Reset timer for each provider
- Minimal UI with dynamic bar icons
-
- Local-first TUI observability for AI coding agent sessions
- Supports Claude Code, Codex CLI, Gemini CLI, Aider, Cursor exports, Hermes, OpenCode, Kimi, and Copilot-style logs
- Tracks cost, tokens, tool failures, latency, anomalies, health, diffs, and CI gates
- Exports JSON, Markdown, and self-contained HTML reports
- Ships as a single Go binary with Homebrew and one-line installer support
-
- Local-first telemetry and memory graph for AI coding agents
- Ingests Claude Code, Codex, Pi, OpenCode, Cursor, and installed skills
- Exposes session search, cost analytics, skill usage, workflow evidence, dashboard, and MCP queries
-
- Visualize Claude Code's LLM interactions
- Runtime behavior and API data analysis
- Interactive log visualization tool
- Understand core agent workflow and architecture
- Context compaction and task/sub-agent analysis
-
- Read, search, and have Claude summarize your Claude Code conversation logs from the CLI
- Reads session
.jsonlfiles from~/.claude/projects/and renders clean markdown - Cross-project keyword search with
findandsearchcommands currentkeyword resolves the session you're in right now (--before-last-compactsupported)- Ships as both a Claude Code plugin and a standalone binary
-
- Claude Code plugin that surfaces context usage, active tools, running agents, and todo progress
- Statusline-style HUD that updates while the agent works
- TypeScript implementation; MIT licensed
-
- Customizable statusline for the Claude Code CLI
- Powerline support, themes, and rich layouts
- TypeScript-driven
-
- Interactive TUI dashboard for token-cost observability
- Shows where Claude Code, Codex, and Cursor tokens go
- TypeScript-driven
-
hamed-elfayome/Claude-Usage-Tracker
- Native macOS menu-bar app for tracking Claude AI usage limits in real time
- Built with Swift and SwiftUI
-
- DevTools-style inspector for Claude Code
- Session logs, tool calls, token usage, sub-agents, and context window
- TypeScript-driven
-
- CLI for tracking token usage across OpenCode, Claude Code, OpenClaw, Codex, Gemini, and Cursor
- Single Rust binary
-
- Annotate and review coding-agent plans and diffs visually
- Share with your team; send feedback to agents with one click
- TypeScript-driven
-
- Warcraft III Peon voice notifications for Claude Code, Codex, and other agents
- Shell-driven; stops you from babysitting the terminal
-
- Minimal plugin that lets Claude Code call you on the phone
- TypeScript-based
- Useful for long-running autonomous runs
Configuration & Templates
-
abhishekray07/claude-md-templates
- CLAUDE.md best practices and templates
- Global, project, and local configurations
- Framework-specific templates (Next.js, Python/FastAPI)
-
shanraisshan/claude-code-best-practice
- Comprehensive best practices guide
- Prompting, planning, and agent strategies
- Commands, skills, hooks, and workflows
-
- Skills, hooks, templates, and engineering guides for AI-agent-friendly projects
/new-projectscaffolding and/adopt-projectfor existing repos- AI-tuned lint rules, hooks reference, and a per-project self-improvement loop
Cost Optimization
-
Sagargupta16/claude-cost-optimizer
- Save 30-60% on Claude API costs
- 6 guides and 15 optimization strategies
- Token estimator and usage analyzer
- 12 CLAUDE.md templates for efficiency
-
nadimtuhin/claude-token-optimizer
- Up to 90% token savings
- Progressive disclosure patterns
- Framework-specific optimization
- 4-file architecture for minimal context
-
- Claude Code skill that rewrites prompts and responses in compressed caveman-style language
- Claims roughly 65% token reduction by stripping function words and filler
- JavaScript implementation; MIT licensed
- Project site at getcaveman.dev
-
AutomateLab-tech/publishing-skills
- Four composable skills for end-to-end SEO blog publishing
- Topic research, writing pipeline, SVG figures, editorial calendar
- Platform-agnostic: Ghost, WordPress, or any static site
- MIT-0 licensed; installs into Claude Code, Cursor, Codex, and 50+ runtimes
Learning & Documentation
Resources for mastering Claude skills and understanding best practices.
Learning & Tutoring Skills
- Li-Evan/Bloom
- One-on-one Socratic tutor skill based on Benjamin Bloom's "2 Sigma" research
- Writes a syllabus, then teaches one lesson at a time, adapting each next lesson to your inline written feedback
- The opposite of "dump the full answer" — guides you to derive it yourself
- Responds in Chinese; skill at
skills/bloom-tutor
Comprehensive Guides
-
FlorianBruniaux/claude-code-ultimate-guide
- 22,000+ lines of documentation
- Beginner to power user progression
- 225 templates and 41 diagrams
- 271 quiz questions for learning validation
-
- Complete guide from beginner to power user
- Setup, commands, workflows, agents, and skills
- Comprehensive MCP integration documentation
- Hooks system, security, and troubleshooting
- Tips, tricks, and best practices
-
- Complete CLI guide auto-updated every 2 days
- Official docs, releases, and changelog integration
- Skills, hooks, MCP, sub-agents, and plugins
- Tool synergies and workflow examples
- Optimized for both humans and AI agents
-
rosmur/claudecode-best-practices
- Evidence-based practices with 58 citations
- Hooks, skills, and automation scripts
- Task queue management with claude-loop
-
- "Bash is all you need" philosophy
- Build a Claude-like agent harness from scratch
- 12 progressive learning sessions
-
Piebald-AI/claude-code-system-prompts
- Complete collection of Claude Code's system prompts
- 18 builtin tool descriptions and sub-agent prompts
- Updated within minutes of each Claude Code release
- Includes changelog across 136+ versions
- Essential for understanding Claude Code internals
-
diet103/claude-code-infrastructure-showcase
- Production-tested infrastructure patterns
- Auto-activating skills via hooks
- Modular skill pattern with 500-line rule
- Specialized agents for complex tasks
- 6 months of iteration, 15-30 min to integrate
-
- 45 tips from basics to advanced techniques
- Custom status line script and system prompt optimization
- Voice input, container workflows, multi-model orchestration
- Includes dx plugin for enhanced developer experience
- Practical strategies for daily Claude Code usage
-
ChrisWiles/claude-code-showcase
- Comprehensive project configuration example
- Hooks, skills, agents, and commands setup
- GitHub Actions workflows for automation
- Best practices and anti-patterns guide
- Complete directory structure reference
-
OneRedOak/claude-code-workflows
- Production workflows from AI-native startup
- Code review, security review, and design review workflows
- Battle-tested configurations since Claude Code launch
- Video tutorials and demos available
- Real-world applied learnings
-
disler/claude-code-hooks-mastery
- Master Claude Code hooks for deterministic control
- Complete hook lifecycle and payload documentation
- Sub-agents, meta-agents, and team-based validation
- Flow control, error codes, and security implementations
- Custom status lines and output styles
-
- Visual, example-driven guide to Claude Code
- Progresses from basic concepts to advanced agent patterns
- Ships copy-paste templates for immediate use
- Hosted at luongnv.com/claude-howto; MIT licensed
-
Windy3f3f3f3f/how-claude-code-works
- Deep dive into Claude Code internals
- Covers architecture, the agent loop, and context engineering
- Bilingual English/Chinese write-up
-
- Long-form analysis of Claude Code's agent-harness skeleton
- In-depth architectural breakdown, structured as a book
- Chinese-language
-
- Chinese tutorial covering Claude Code, OpenClaw, and Codex
- 36 complete tutorials, three learning paths from beginner to enterprise
- Markdown-only bundle
-
jamesmurdza/awesome-ai-devtools
- Curated list of AI-powered developer tools
- Useful adjacent reading when shopping for Claude Code integrations
-
Meirtz/Awesome-Context-Engineering
- Comprehensive survey on context engineering
- From prompt engineering to production-grade AI systems
- Hundreds of papers and frameworks referenced
-
KhazP/vibe-coding-prompt-template
- Templates and workflow for generating PRDs, tech designs, and MVPs
- Designed for use with AI IDEs and Claude Code
- Markdown-only bundle
Skill Development
Tools and frameworks for creating your own Claude skills.
-
alirezarezvani/claude-code-skill-factory
- Complete toolkit for building production-ready skills
- Deployment automation
- Best practices and templates
- Scale from prototype to production
-
- Open specification, documentation, and reference SDK for Agent Skills
- Defines the
SKILL.mdformat used across Claude Code and compatible agents - Maintained by Anthropic with community contributions
- Apache 2.0 licensed (code) and CC-BY-4.0 (docs); companion site at agentskills.io
-
- Universal
SKILL.mdloader installable vianpm i -g openskills - Brings Anthropic's skills system to Claude Code, Cursor, Windsurf, Aider, Codex, and any tool that reads
AGENTS.md - Project-local installs default to
./.claude/skills;--globalwrites to~/.claude/skills - Supports private git repos and local paths
- Universal
-
- Converts documentation sites, GitHub repositories, and PDFs into Claude AI skills
- Automatic conflict detection between overlapping skill definitions
- AST parsing, OCR for PDFs, and web scraping pipelines
- MIT licensed; hosted version at skillseekersweb.com
-
- Open-source agent-skills builder
- Define skills by workflow and run on Claude Code, Cursor, Codex, and others
- TypeScript-driven
-
- Microsoft's Agent Package Manager
- Distributes agent skills, MCP servers, and custom agents
- Python-driven
-
- Secure, validated skill registry for professional AI coding agents
- Extends Antigravity, Claude Code, Cursor, and Copilot
- TypeScript-driven
-
- Spec-driven development harness for Claude Code
- Turns approved specs into long-running autonomous implementation
- TypeScript-driven, minimal and adaptable
Contributing
We welcome contributions! To add a skill or tool to this list:
- Ensure the project is actively maintained
- Verify it provides clear documentation
- Check that it offers real value to Claude users
- Submit a pull request with a clear description
Quality over quantity - we prioritize well-documented, production-ready skills over experimental projects.
License
CC0 - Public Domain
This curated list is maintained by the community for the community.
- Septim Agents Pack - 15 named Claude Code sub-agents (Atlas, Luca, Canon, Ember, Tally, Nova, Ward, Mira, Juno, Pip, Hart, Halo, Beacon, Loom, Lynx) covering the full exec layer. Drops into
~/.claude/agents/. Open-source sample at github.com/septimlabs-code/septim-agents-pack-sample. $49 lifetime.