179 skills · 35 agents · 150 hooks · 33 modular plugins
Why OrchestKit? · Quick Start · Commands · Skills · Agents · FAQ
# Install in 30 seconds
/plugin marketplace add yonatangross/orchestkit
/plugin install ork
# Verify installation
/ork:doctorThat's it. Skills auto-activate based on your prompts. Hooks protect your code automatically.
The Problem: Every Claude Code session starts from zero. You explain your stack, your patterns, your preferences—again and again.
The Solution: OrchestKit gives Claude persistent knowledge of 163 production patterns, 34 specialized agents, and 147 security/quality hooks that work automatically.
|
Without OrchestKit |
With OrchestKit |
YOUR PROMPT
│
┌─────────────────────┼─────────────────────┐
│ │ │
▼ ▼ ▼
┌───────────────┐ ┌───────────────┐ ┌───────────────┐
│ 🛡️ HOOKS │ │ 📚 SKILLS │ │ 🤖 AGENTS │
│ │ │ │ │ │
│ Security gate │ │ Pattern libs │ │ Specialists │
│ Git protect │ │ Best practice │ │ Auto-activate │
│ Quality check │ │ Code templates│ │ Domain expert │
│ │ │ │ │ │
│ 150 hooks │ │ 179 skills │ │ 35 agents │
└───────┬───────┘ └───────┬───────┘ └───────┬───────┘
│ │ │
│ ┌───────────────┴───────────────┐ │
│ │ │ │
▼ ▼ ▼ ▼
┌─────────────────────────────────────────────────────┐
│ │
│ ✅ PRODUCTION-READY CODE │
│ │
│ • Follows your stack's patterns │
│ • Security validated │
│ • Tests included │
│ • Ready to commit │
│ │
└─────────────────────────────────────────────────────┘
flowchart LR
subgraph Trigger["⚡ TRIGGER"]
P[Your Prompt]
end
subgraph Parallel["⚙️ PARALLEL PROCESSING"]
direction TB
H["🛡️ Hooks<br/>Security & Quality"]
S["📚 Skills<br/>Pattern Injection"]
A["🤖 Agents<br/>Auto-Activation"]
end
subgraph Execute["🚀 EXECUTE"]
direction TB
V[Validate]
G[Generate]
T[Test]
end
subgraph Output["✅ OUTPUT"]
C[Production Code]
end
P --> H & S & A
H --> V
S --> G
A --> G
V --> T
G --> T
T --> C
classDef trigger fill:#e3f2fd,stroke:#1976d2,stroke-width:2px
classDef hooks fill:#ffebee,stroke:#c62828,stroke-width:2px
classDef skills fill:#e8f5e9,stroke:#2e7d32,stroke-width:2px
classDef agents fill:#fff3e0,stroke:#ef6c00,stroke-width:2px
classDef execute fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px
classDef output fill:#e0f2f1,stroke:#00695c,stroke-width:2px
class P trigger
class H hooks
class S skills
class A agents
class V,G,T execute
class C output
# Commit with automatic validation
/ork:commit✓ Pre-commit checks passed
✓ Tests: 42 passed
✓ Security scan: clean
✓ Commit created: feat(api): add user endpoint
# Verify your implementation works
/ork:verifyRunning 3 verification agents in parallel...
├─ test-generator: 4 tests added, coverage 67% → 82%
├─ security-auditor: No vulnerabilities found
└─ code-quality-reviewer: LGTM, minor suggestions noted
# Explore unfamiliar codebase
/ork:exploreCodebase Analysis:
├─ Backend: FastAPI + SQLAlchemy 2.0 (12 endpoints)
├─ Frontend: React 19 + Vite (23 components)
├─ Database: PostgreSQL + pgvector
└─ Key patterns: Cursor pagination, JWT auth, SSE events
33 domain-specific plugins — install only what you need:
┌─────────────────────────────────────────────────────────────────────────────┐
│ PLUGIN ARCHITECTURE │
├─────────────────────────────────────────────────────────────────────────────┤
│ │
│ ┌─────────────────────────────────────────────────────────────────────┐ │
│ │ orchestkit-complete │ │
│ │ Full toolkit: all 33 plugins │ │
│ └─────────────────────────────────────────────────────────────────────┘ │
│ │ │
│ ┌───────────────────────┼───────────────────────┐ │
│ ▼ ▼ ▼ │
│ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐ │
│ │ 🧠 AI/LLM │ │ ⚡ Backend │ │ 🎨 Frontend │ │
│ │ 7 plugins │ │ 6 plugins │ │ 4 plugins │ │
│ │ │ │ │ │ │ │
│ │ • ork-rag │ │ • ork-fastapi │ │ • ork-react-core│ │
│ │ • ork-langgraph │ │ • ork-database │ │ • ork-ui-design │ │
│ │ • ork-llm-* │ │ • ork-async │ │ • ork-frontend-*│ │
│ └─────────────────┘ └─────────────────┘ └─────────────────┘ │
│ │
│ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐ │
│ │ 🧪 Testing │ │ 🔒 Security │ │ 🚀 DevOps │ │
│ │ 2 plugins │ │ 1 plugin │ │ 3 plugins │ │
│ │ │ │ │ │ │ │
│ │ • ork-testing-* │ │ • ork-security │ │ • ork-cicd │ │
│ │ • ork-e2e │ │ │ │ • ork-git │ │
│ └─────────────────┘ └─────────────────┘ └─────────────────┘ │
│ │
└─────────────────────────────────────────────────────────────────────────────┘
# Full toolkit (recommended for new users)
/plugin install ork@orchestkit
# Individual domains (advanced)
/plugin install ork-core@orchestkit # Core foundation (required)
/plugin install ork-rag@orchestkit # RAG & retrieval
/plugin install ork-fastapi@orchestkit # FastAPI backend
/plugin install ork-react-core@orchestkit # React frontend
/plugin install ork-testing-core@orchestkit # Testing patterns📦 All 33 Plugins by Category
| Category | Plugins | Description |
|---|---|---|
| Core | ork-core, ork-context, ork-memory |
Foundation, context management, persistence |
| AI/LLM | ork-rag, ork-rag-advanced, ork-langgraph-core, ork-langgraph-advanced, ork-llm-core, ork-llm-advanced, ork-ai-observability |
RAG, agents, LLM patterns (7 plugins) |
| Backend | ork-fastapi, ork-database, ork-async, ork-architecture, ork-backend-advanced, ork-graphql |
APIs, databases, async (6 plugins) |
| Frontend | ork-react-core, ork-ui-design, ork-frontend-performance, ork-frontend-advanced |
React, UI, performance (4 plugins) |
| Testing | ork-testing-core, ork-testing-e2e |
Unit, integration, E2E (2 plugins) |
| Security | ork-security |
OWASP, auth, validation, guardrails |
| DevOps | ork-cicd, ork-infrastructure, ork-git |
CI/CD, infra, git workflows (3 plugins) |
| Data | ork-data-engineering, ork-evaluation |
ETL, embeddings, golden datasets (2 plugins) |
| Workflows | ork-workflows-core, ork-workflows-advanced |
Implementation, verification flows (2 plugins) |
| Other | ork-accessibility, ork-mcp, ork-product |
A11y, MCP servers, product management (3 plugins) |
21 slash commands organized by workflow:
| Command | Description |
|---|---|
/ork:commit |
Conventional commit with pre-commit checks |
/ork:create-pr |
Create PR with summary and test plan |
/ork:review-pr |
Code review checklist |
/ork:git-recovery-command |
Recover from git mistakes |
| Command | Description |
|---|---|
/ork:remember |
Save information to persistent memory |
/ork:recall |
Retrieve from memory |
/ork:load-context |
Load relevant memories at session start |
/ork:mem0-sync |
Sync memories to Mem0 cloud |
| Command | Description |
|---|---|
/ork:explore |
Analyze codebase structure |
/ork:implement |
Implement feature with agent guidance |
/ork:verify |
Verify implementation correctness |
/ork:fix-issue |
Fix a GitHub issue |
| Command | Description |
|---|---|
/ork:doctor |
Check plugin health |
/ork:configure |
Setup MCP servers |
| Command | Description |
|---|---|
/ork:brainstorming |
Structured ideation session |
/ork:feedback |
Submit feedback or suggestions |
/ork:add-golden |
Add golden test dataset |
/ork:skill-evolution |
Evolve skills based on usage |
/ork:worktree-coordination |
Coordinate multiple Claude instances |
179 skills with progressive loading (~70% token savings):
| Category | Count | Key Skills |
|---|---|---|
| RAG & Retrieval | 6 | rag-retrieval, contextual-retrieval, reranking-patterns, hyde-retrieval, query-decomposition, agentic-rag-patterns |
| LLM Patterns | 8 | function-calling, llm-streaming, llm-evaluation, prompt-engineering-suite, fine-tuning-customization, vision-language-models, high-performance-inference, semantic-caching |
| Agents & Orchestration | 7 | agent-loops, multi-agent-orchestration, langgraph-* (179 skills), alternative-agent-frameworks |
| Safety & Security | 6 | llm-safety-patterns, advanced-guardrails, mcp-security-hardening, llm-testing |
| Category | Count | Key Skills |
|---|---|---|
| FastAPI & Async | 4 | fastapi-advanced, asyncio-advanced, sqlalchemy-2-async, connection-pooling |
| Task Processing | 3 | celery-advanced, temporal-io, background-jobs |
| APIs & Communication | 3 | strawberry-graphql, grpc-python, streaming-api-patterns |
| Architecture | 5 | saga-patterns, cqrs-patterns, event-sourcing, outbox-pattern, aggregate-patterns |
| Resilience | 4 | rate-limiting, idempotency-patterns, distributed-locks, resilience-patterns |
| Category | Count | Key Skills |
|---|---|---|
| React & State | 6 | react-server-components-framework, zustand-patterns, tanstack-query-advanced, form-state-patterns |
| Performance | 5 | core-web-vitals, lazy-loading-patterns, image-optimization, render-optimization |
| UI & Animation | 6 | view-transitions, scroll-driven-animations, motion-animation-patterns, radix-primitives, shadcn-patterns |
| Data Viz & PWA | 4 | recharts-patterns, dashboard-patterns, pwa-patterns, responsive-patterns |
| Build & Quality | 2 | vite-advanced, biome-linting |
| Category | Count | Key Skills |
|---|---|---|
| Unit & Integration | 4 | pytest-advanced, unit-testing, integration-testing, msw-mocking |
| Advanced Testing | 4 | property-based-testing, contract-testing, e2e-testing, vcr-http-recording |
| Test Data | 2 | test-data-management, golden-dataset-* (179 skills) |
owasp-top-10 · auth-patterns · input-validation · defense-in-depth · security-scanning
github-operations · git-workflow · stacked-prs · release-management · observability-monitoring · devops-deployment · zero-downtime-migration · database-versioning · alembic-migrations
34 specialized agents organized by domain:
| Agent | Specialty |
|---|---|
backend-system-architect |
REST/GraphQL APIs, microservices, clean architecture |
database-engineer |
PostgreSQL, pgvector, schema design, migrations |
event-driven-architect |
Event sourcing, CQRS, message queues |
data-pipeline-engineer |
ETL, data flows, batch processing |
python-performance-engineer |
Async optimization, profiling, caching |
infrastructure-architect |
Cloud architecture, scaling patterns |
| Agent | Specialty |
|---|---|
frontend-ui-developer |
React 19, TypeScript, component architecture |
rapid-ui-designer |
Quick prototypes, design systems |
performance-engineer |
Core Web Vitals, bundle optimization |
accessibility-specialist |
WCAG 2.2, ARIA, keyboard navigation |
ux-researcher |
User flows, usability analysis |
| Agent | Specialty |
|---|---|
llm-integrator |
LLM APIs, prompt design, token optimization |
workflow-architect |
LangGraph, multi-agent orchestration |
ai-safety-auditor |
Guardrails, prompt injection defense |
prompt-engineer |
Chain-of-thought, few-shot learning |
multimodal-specialist |
Vision, audio, multi-modal pipelines |
| Agent | Specialty |
|---|---|
security-auditor |
OWASP Top 10, vulnerability assessment |
security-layer-auditor |
Defense-in-depth, authentication flows |
| Agent | Specialty |
|---|---|
test-generator |
Unit/integration tests, MSW, coverage |
code-quality-reviewer |
Code review, best practices, refactoring |
system-design-reviewer |
Architecture review, trade-offs |
debug-investigator |
Root cause analysis, debugging |
| Agent | Specialty |
|---|---|
ci-cd-engineer |
GitHub Actions, deployment pipelines |
deployment-manager |
Release coordination, rollback strategies |
release-engineer |
Versioning, changelogs, release automation |
git-operations-engineer |
Branch strategies, merge workflows |
monitoring-engineer |
Prometheus, Grafana, alerting |
metrics-architect |
Observability, KPIs, dashboards |
| Agent | Specialty |
|---|---|
product-strategist |
Feature prioritization, roadmaps |
business-case-builder |
ROI analysis, business justification |
market-intelligence |
Competitive analysis, trends |
prioritization-analyst |
Backlog management, impact scoring |
requirements-translator |
Specs to implementation plans |
| Agent | Specialty |
|---|---|
documentation-specialist |
API docs, READMEs, technical writing |
flowchart TB
subgraph Input["📝 INPUT"]
P["Your Prompt"]
end
subgraph OrchestKit["🔷 ORCHESTKIT PLUGIN"]
direction TB
subgraph Hooks["🛡️ 147 HOOKS"]
direction LR
H1["PreToolUse"]
H2["PostToolUse"]
H3["Permission"]
H4["Lifecycle"]
end
subgraph Skills["📚 161 SKILLS"]
direction LR
S1["Backend"]
S2["Frontend"]
S3["AI/ML"]
S4["Testing"]
end
subgraph Agents["🤖 34 AGENTS"]
direction LR
A1["Architects"]
A2["Engineers"]
A3["Reviewers"]
A4["Specialists"]
end
end
subgraph Output["✅ OUTPUT"]
C["Production Code"]
end
P --> Hooks
P --> Skills
P --> Agents
Hooks --> C
Skills --> C
Agents --> C
classDef input fill:#e3f2fd,stroke:#1976d2,stroke-width:2px,color:#0d47a1
classDef hooks fill:#ffebee,stroke:#c62828,stroke-width:2px,color:#b71c1c
classDef skills fill:#e8f5e9,stroke:#2e7d32,stroke-width:2px,color:#1b5e20
classDef agents fill:#fff3e0,stroke:#ef6c00,stroke-width:2px,color:#e65100
classDef output fill:#e0f2f1,stroke:#00695c,stroke-width:2px,color:#004d40
classDef container fill:#fafafa,stroke:#9e9e9e,stroke-width:1px
class P input
class H1,H2,H3,H0 hooks
class S1,S2,S3,S179 skills
class A1,A2,A3,A35 agents
class C output
class OrchestKit container
orchestkit/
├── .claude-plugin/
│ └── marketplace.json # 33 modular plugins
├── plugins/ # Modular plugin bundles
│ └── ork-<domain>/ # Domain-specific plugin
│ ├── .claude-plugin/
│ │ └── plugin.json # Plugin manifest
│ ├── commands/ # Slash commands
│ ├── agents/ # Specialized agents
│ ├── skills/ # Knowledge modules
│ └── scripts/ # Hook executables
├── skills/ # 161 knowledge modules (full)
│ └── <skill-name>/
│ ├── SKILL.md # Overview + patterns (~500 tokens)
│ ├── references/ # Deep-dive guides (~200 tokens)
│ ├── scripts/ # Executable code and generators
│ └── assets/ # Templates and copyable files
├── agents/ # 34 specialized agents
├── commands/ # 21 slash commands
├── hooks/ # 147 lifecycle hooks
└── tests/ # Validation suite
| Feature | OrchestKit | claude-code-showcase | DIY Hooks |
|---|---|---|---|
| Skills/Patterns | ✅ 161 | ❌ 0 | |
| Specialized Agents | ✅ 34 | ❌ 0 | |
| Security Layers | ✅ 8-layer | ❌ Manual | |
| AI/ML Patterns | ✅ 27 | ❌ None | |
| Testing Patterns | ✅ 10 | ❌ None | |
| Setup Time | ✅ 2 min | ❌ Hours | |
| Maintenance | ✅ Auto | ❌ Manual | ❌ Manual |
| Progressive Loading | ✅ Yes | ❌ No | ❌ No |
| Memory Integration | ✅ Graph + Mem0 | ❌ None | ❌ None |
/ork:configure| Server | Purpose | When Active |
|---|---|---|
| Context7 | Up-to-date library docs | ✅ Until 75% context |
| Memory | Knowledge graph (PRIMARY) | ✅ Until 90% context |
| Sequential Thinking | Complex reasoning | ✅ Until 60% context |
| Mem0 | Semantic search (optional) | ⚙️ Requires API key |
Browser Automation: Use agent-browser CLI via Bash (93% less context than MCP).
CLAUDE_PROJECT_DIR # Your project directory
CLAUDE_PLUGIN_ROOT # Plugin installation path
CLAUDE_SESSION_ID # Current session ID
MEM0_API_KEY # Optional: Mem0 cloud integration❓ Plugin not found after installation?
# Verify installation
/plugin list
# Reinstall if needed
/plugin uninstall ork
/plugin marketplace add yonatangross/orchestkit
/plugin install ork❓ Plugin works locally but not in other projects?
The plugin may be installed with local scope (tied to one project). To fix:
# Check current scope
cat ~/.claude/plugins/installed_plugins.json | grep -A5 "ork@orchestkit"If you see "scope": "local" or "scope": "project", change it to "scope": "user" and remove the "projectPath" field. Then restart Claude Code.
❓ Plugin installed but not showing in /plugin list?
The plugin may be installed but not enabled. Check ~/.claude/settings.json:
cat ~/.claude/settings.json | grep orkIf ork@orchestkit is missing from enabledPlugins, add it:
"enabledPlugins": {
"ork@orchestkit": true,
...
}Then restart Claude Code.
❓ Old version loading despite update?
The plugin cache may have a stale version. Clear and reinstall:
# Check cached version
ls ~/.claude/plugins/cache/orchestkit/ork/
# Clear cache and reinstall
/plugin uninstall ork
rm -rf ~/.claude/plugins/cache/orchestkit/ork/
/plugin install ork❓ Hooks not firing?
- Check hook logs:
tail -f hooks/logs/*.log - Verify settings: Check
.claude/settings.jsonexists - Run diagnostics:
/ork:doctor
❓ How do I add my own skills?
mkdir -p skills/my-skill/references
cat > skills/my-skill/SKILL.md << 'EOF'
---
name: my-skill
description: What this skill provides
tags: [keyword1, keyword2]
---
# My Skill
Overview of patterns...
EOF
./tests/skills/structure/test-skill-md.sh❓ Works with existing projects?
Yes! OrchestKit is additive—it won't modify your files. Skills and agents activate automatically based on context.
❓ How much context does this use?
Progressive loading minimizes usage:
| Stage | Tokens | When |
|---|---|---|
| Discovery | ~50 | Always |
| Overview | ~500 | Skill relevant |
| Specific | ~200 | Implementing |
| Templates | ~300 | Generating |
Result: ~70% savings vs loading everything.
❓ Claude Code version requirements?
Requires Claude Code ≥2.1.16 for full features:
- CC 2.1.6: Agent skill injection
- CC 2.1.7: Parallel hook execution
- CC 2.1.9: additionalContext injection
- CC 2.1.11: Setup hooks
- CC 2.1.14: Plugin versioning (git SHA/tag pinning)
- CC 2.1.15: Engine field, plugin search
- CC 2.1.16: Task dependencies, VSCode native plugins
OrchestKit uses a build system to assemble modular plugins from source files.
src/
├── skills/ # 179 skills (single source of truth - edit these)
├── agents/ # 35 agents (single source of truth - edit these)
└── hooks/ # 150 hooks (single source of truth - edit these)
manifests/ # Plugin definitions (34 manifests - edit these)
├── ork.json
├── ork-rag.json
└── ...
scripts/
└── build-plugins.sh # Assembles plugins from src/ + manifests/
plugins/ # Generated (DO NOT EDIT - created by build script)
├── ork/
├── ork-rag/
└── ...
.claude-plugin/
└── marketplace.json # Generated marketplace manifest
Important: Always edit files in src/ and manifests/, not plugins/. The plugins/ directory is regenerated on every build.
# Build all plugins (two methods)
npm run build
# OR
bash scripts/build-plugins.sh
# Test local installation
/plugin marketplace add file://$(pwd)
/plugin install ork-rag
# Verify
/ork:doctor- Edit source files in
src/skills/,src/agents/, ormanifests/ - Run
npm run buildto regenerateplugins/ - Test changes locally
- Commit changes (only
src/andmanifests/are tracked in git)
# Run all tests
npm test
# OR
./tests/run-all-tests.sh
# Individual test suites
npm run test:security # Security tests (must pass)
npm run test:skills # Validate skill structure
npm run test:agents # Validate agent frontmatter
# Or use bash directly
./tests/security/run-security-tests.sh
./tests/skills/test-skill-structure.sh
./tests/agents/test-agent-frontmatter.sh- Fork → 2. Branch → 3. Test → 4. PR
See CONTRIBUTING.md for guidelines.
v4.28.x — CC 2.1.16 support, README marketplace alignment, subagent compatibility
Recent:
- CC 2.1.16: Task dependencies, VSCode native plugins
- agent-browser CLI integration (93% less context vs Playwright MCP)
- 33 modular plugins for selective installation
- AI safety skills (guardrails, prompt injection defense)
MIT License — see LICENSE
Documentation · Issues · Discussions
Built with Claude Code · Maintained by @yonatangross