One install. Full power. From idea to production — without leaving Claude Code.
Quick Install · What's Inside · Agents · Commands · Skills · MCPs · Workflow · Templates · Docs
Claude OS is a production-ready, fullstack AI development system built on top of Claude Code. It transforms Claude Code into a complete development environment with specialized agents, 64+ commands, 69+ skills, multi-agent orchestration, context-rot prevention, SEO automation, security scanning, and built-in best practices for every layer of your stack.
Stop configuring. Start building.
curl -fsSL https://raw.githubusercontent.com/royaluniondesign-sys/claude-os/main/install.sh | bash| Layer | Tools | Coverage |
|---|---|---|
| Project Lifecycle | GSD v1.22 | Idea → Specs → Architecture → Build → Verify → Ship |
| Multi-Agent Orchestration | ECC v1.7 + GSD | Parallel agents, wave execution, subagent spawning |
| Code Quality | Rules + Reviewers | TypeScript, Python, Go, Java — enforced automatically |
| Security | AgentShield (1282 tests, 102 rules) | OWASP, secrets, dependency audit |
| SEO | claude-seo + mcp-gsc | Technical SEO, GSC integration, Core Web Vitals |
| Testing | TDD + E2E + Coverage | Unit, integration, e2e, performance |
| Memory | Session hooks | Context persists across sessions automatically |
| System Control | DesktopCommanderMCP | Terminal, filesystem, processes, Excel, PDF |
| Continuous Learning | Instinct engine | Claude learns your patterns and improves over time |
curl -fsSL https://raw.githubusercontent.com/royaluniondesign-sys/claude-os/main/install.sh | bashOr manually:
git clone https://github.com/royaluniondesign-sys/claude-os.git
cd claude-os
./install.shRequirements: Claude Code, Node.js 18+, Python 3.9+, Git
Claude OS follows a proven spec-driven development cycle that eliminates context rot and delivers production-quality output consistently:
┌─────────────────────────────────────────────────────────────────┐
│ CLAUDE OS CYCLE │
├─────────────────────────────────────────────────────────────────┤
│ │
│ /gsd:new-project → Questions → Research → Specs → Roadmap │
│ /gsd:discuss-phase → Lock decisions before planning │
│ /gsd:plan-phase → Research + atomic XML plans │
│ /gsd:execute-phase → Parallel agents, 200k tokens each │
│ /gsd:verify-work → Human + automated verification │
│ /gsd:complete-milestone → Archive, tag, release │
│ │
│ Each plan runs in a FRESH context window. │
│ No degradation. No "I'll be more concise now." │
│ │
└─────────────────────────────────────────────────────────────────┘
31 specialized agents ready to spawn on demand:
| Agent | Role |
|---|---|
gsd-planner |
Creates atomic XML implementation plans |
gsd-executor |
Executes tasks in isolated 200k token contexts |
gsd-verifier |
Validates deliverables against phase goals |
gsd-debugger |
Diagnoses failures systematically |
gsd-roadmapper |
Builds phased project roadmaps |
gsd-phase-researcher |
Deep research before planning |
gsd-plan-checker |
Verifies plans before execution |
gsd-codebase-mapper |
Analyzes existing codebases |
gsd-project-researcher |
Domain and ecosystem research |
gsd-research-synthesizer |
Synthesizes research findings |
gsd-integration-checker |
Validates cross-component integration |
| Agent | Role |
|---|---|
architect |
System design, stack selection, architecture decisions |
chief-of-staff |
Orchestrates multi-agent workflows |
planner |
Task breakdown and planning |
code-reviewer |
Code quality, patterns, readability |
security-reviewer |
OWASP compliance, vulnerability detection |
database-reviewer |
Schema design, query optimization |
tdd-guide |
Test-driven development guidance |
refactor-cleaner |
Code cleanup and refactoring |
doc-updater |
Documentation synchronization |
e2e-runner |
End-to-end test execution |
build-error-resolver |
Diagnoses and fixes build failures |
python-reviewer |
Python-specific code review |
go-reviewer |
Go-specific code review |
go-build-resolver |
Go build error diagnosis |
| Agent | Role |
|---|---|
seo-technical |
Technical SEO audit |
seo-content |
Content optimization |
seo-performance |
Core Web Vitals, page speed |
seo-schema |
Structured data / JSON-LD |
seo-sitemap |
Sitemap generation and optimization |
seo-visual |
Visual SEO analysis |
64 commands across two systems:
/gsd:new-project # Full project init: questions → research → specs → roadmap
/gsd:map-codebase # Analyze existing codebase before starting
/gsd:discuss-phase [N] # Capture implementation decisions
/gsd:plan-phase [N] # Research + create atomic plans
/gsd:execute-phase <N> # Execute in parallel waves
/gsd:verify-work [N] # Human acceptance testing
/gsd:quick # Ad-hoc tasks with GSD guarantees
/gsd:debug [desc] # Systematic debugging
/gsd:progress # Where am I? What's next?
/gsd:new-milestone # Start next version
/gsd:complete-milestone # Archive and tag release
/gsd:settings # Configure model profiles
/gsd:health --repair # Diagnose and repair installation/orchestrate # Spawn and coordinate multiple agents
/multi-plan # Parallel planning with multiple agents
/multi-execute # Parallel execution across agents
/multi-frontend # Frontend-focused multi-agent workflow
/multi-backend # Backend-focused multi-agent workflow
/multi-workflow # Full-stack multi-agent orchestration
/code-review # Comprehensive code review
/security-scan # AgentShield security scan (102 rules)
/tdd # Test-driven development workflow
/e2e # End-to-end testing
/test-coverage # Coverage analysis
/verify # Verification loop
/checkpoint # Save progress checkpoint
/learn # Extract patterns into reusable skills
/evolve # Evolve existing skills
/refactor-clean # Automated refactoring
/update-docs # Sync documentation
/plan # Structured planning
/pm2 # Multi-service process management69 production-ready skills covering every part of the stack:
frontend-patterns frontend-slides liquid-glass-design swiftui-patterns swift-concurrency-6-2 swift-actor-persistence swift-protocol-di-testing
backend-patterns api-design python-patterns golang-patterns django-patterns springboot-patterns jpa-patterns
postgres-patterns database-migrations clickhouse-io
docker-patterns deployment-patterns
tdd-workflow e2e-testing python-testing golang-testing django-tdd springboot-tdd django-verification springboot-verification cpp-testing eval-harness verification-loop
security-review security-scan django-security springboot-security
seo seo-audit seo-technical seo-content seo-page seo-plan seo-schema seo-sitemap seo-images seo-geo seo-hreflang seo-programmatic seo-competitor-pages
foundation-models-on-device iterative-retrieval cost-aware-llm-pipeline regex-vs-llm-structured-text content-hash-cache-pattern
article-writing content-engine market-research investor-materials investor-outreach strategic-compact
coding-standards cpp-coding-standards java-coding-standards search-first continuous-learning continuous-learning-v2 skill-stocktake skill-create configure-ecc project-guidelines-example
Model Context Protocol servers that extend Claude's capabilities:
| MCP | What it gives Claude |
|---|---|
desktop-commander |
Full terminal control, filesystem ops, process management, Excel/PDF support |
github (optional) |
PR creation, issue management, CI/CD integration |
postgresql (optional) |
Direct database access, schema inspection, query execution |
gsc (optional) |
Google Search Console data, SEO insights |
# GitHub integration
./scripts/install-mcp-github.sh
# PostgreSQL
./scripts/install-mcp-postgres.sh
# Google Search Console (SEO)
./scripts/install-mcp-gsc.shLanguage-specific rules enforced automatically on every task:
rules/
├── common/ # Universal rules: security, naming, structure
├── typescript/ # TS-specific: types, async patterns, imports
├── python/ # PEP 8, type hints, dependency management
└── golang/ # Go idioms, error handling, concurrency
Starter templates with Claude OS pre-configured:
| Template | Stack | Includes |
|---|---|---|
nextjs-fullstack |
Next.js 15 + PostgreSQL + Auth | App Router, tRPC, Drizzle, Tailwind |
api-rest |
Node.js / Python / Go | OpenAPI spec, auth, rate limiting, tests |
shopify-theme |
Shopify + Liquid | Theme architecture, performance, SEO |
saas-starter |
Next.js + Stripe + Auth | Billing, orgs, roles, dashboards |
# Use a template
./scripts/new-project.sh nextjs-fullstack my-appBalance quality vs cost per task:
/gsd:set-profile quality # Opus everywhere (best results)
/gsd:set-profile balanced # Opus planning + Sonnet execution (default)
/gsd:set-profile budget # Sonnet planning + Haiku verificationAdd to .claude/settings.json in your project:
{
"permissions": {
"allow": [
"Bash(git:*)",
"Bash(npm:*)",
"Bash(npx:*)"
],
"deny": [
"Read(.env)",
"Read(.env.*)",
"Read(**/*.pem)",
"Read(**/*.key)"
]
}
}~/.claude/
├── agents/ (31 agents)
├── commands/ (64 commands)
│ └── gsd/ (31 GSD commands)
├── skills/ (69 skills)
├── rules/ (4 language rulesets)
├── hooks/ (session memory + context monitor)
├── hooks.json (hook configuration)
├── get-shit-done/ (GSD templates and config)
└── settings.json (updated with hooks + statusline)
Claude OS aggregates and builds on the work of an incredible open source community:
| Project | Contribution |
|---|---|
| affaan-m/everything-claude-code | Core agents, skills, commands, rules, hooks |
| gsd-build/get-shit-done | Project lifecycle system |
| wonderwhy-er/DesktopCommanderMCP | System control MCP |
| AgriciDaniel/claude-seo | SEO skill suite |
| davila7/claude-code-templates | Template components |
| anthropics/skills | Official Anthropic skills |
All original licenses are preserved. See ATTRIBUTION.md for details.
To contribute:
- Fork this repo
- Add your agent / skill / command / MCP in the right folder
- Document it in this README
- Open a PR
- GitHub Actions MCP integration
- Terraform / infrastructure agent
- Monitoring agent (Datadog / Grafana)
- n8n workflow integration
- One-click Vercel / Railway deployment templates
- Web UI for skill/agent management
MIT — see LICENSE
Built for developers who ship.
