Raffles IT Kit

AI Agent Enhancement Toolkit — specialist agents, skills, and workflows for Claude Code, Cursor, and Windsurf.

Quick Install

npx create-raffles-it

Or install globally:

npm install -g create-raffles-it
create-raffles-it init

This copies the full agent toolkit — agents/, skills/, workflows/, configs/, prompts/, and rules/ — directly into your project root.

What's Included

| Component | Count | Description | | ------------- | ----- | -------------------------------------------------------------------- | | Agents | 19 | Specialist AI personas (frontend, backend, security, PM, QA, etc.) | | Skills | 19 | Modular domain-specific knowledge packs, loaded on demand | | Workflows | 11 | Slash command procedures for complex multi-step tasks |

How to Use

1. Install the kit

# Run inside your project root
npx create-raffles-it

2. Open in your AI editor

# Claude Code
claude .

# Or open with Cursor / Windsurf — agents are auto-detected from the workspace

3. Just describe what you need — no configuration required

Agents are automatically selected based on your request. You never need to pick one manually.

You: "Create a responsive product card with Tailwind CSS and dark mode"
AI: 🤖 Routing to frontend-specialist
    Loading skills: react-best-practices, tailwind-patterns, frontend-design

You: "Build a JWT auth API with refresh tokens and rate limiting"
AI: 🤖 Routing to backend-specialist
    Loading skills: api-patterns, nodejs-best-practices, clean-code

You: "Our login endpoint returns 500 only in production"
AI: 🤖 Routing to debugger
    Loading skills: systematic-debugging

You: "Help me prioritize the backlog for our MVP launch in 3 weeks"
AI: 🤖 Routing to product-owner
    Loading skills: plan-writing, brainstorming

Agents

| Agent | Triggers on… | | ------------------------ | --------------------------------------- | | orchestrator | orchestrate, coordinate, multi-step | | frontend-specialist | component, react, UI, CSS, tailwind | | backend-specialist | backend, server, API, endpoint, auth | | database-architect | database, schema, migration, SQL | | debugger | bug, error, crash, not working, fix | | devops-engineer | deploy, docker, CI/CD, release | | security-auditor | security, vulnerability, OWASP, XSS | | penetration-tester | pentest, exploit, red team, offensive | | test-engineer | test, spec, coverage, jest, playwright | | qa-automation-engineer | e2e, automated test, regression | | performance-optimizer | performance, speed, lighthouse, memory | | explorer-agent | explore, audit, analyse repo, map | | code-archaeologist | legacy, refactor, reverse engineer | | project-planner | plan, roadmap, breakdown, milestones | | product-manager | requirements, user story, specs | | product-owner | backlog, MVP, PRD, stakeholder | | documentation-writer | write docs, README, changelog | | seo-specialist | SEO, meta, sitemap, core web vitals |

Slash Command Workflows

Invoke with /command in Claude Code, Cursor, or Windsurf chat:

| Command | Description | Example | | ---------------- | ---------------------------------------------------- | ------------------------------------------------ | | /create | Scaffold a new feature end-to-end | /create user auth with Google OAuth | | /debug | Structured root-cause analysis | /debug payment webhook failing in production | | /plan | Break a large task into a sprint plan | /plan migrate monolith to microservices | | /deploy | Pre-flight checks → build → deploy → smoke test | /deploy to production with zero downtime | | /test | Generate unit, integration, and E2E tests | /test the auth module | | /enhance | Improve performance, readability, type safety | /enhance the user service class | | /brainstorm | Socratic discovery for architecture or product ideas | /brainstorm architecture for real-time chat | | /orchestrate | Coordinate multiple agents in parallel | /orchestrate build a full SaaS billing system | | /review | Code review: OWASP, performance, conventions | /review src/api/payments.ts | | /status | Project health: tests, lint, TODOs, security flags | /status | | /ui-ux-pro-max | Design with 50 UI styles, 21 palettes, 50 fonts | /ui-ux-pro-max redesign the dashboard |

Skills

Skills are modular knowledge packs. Each agent loads its relevant skills automatically.

| Category | Skills | | ------------------- | -------------------------------------------------------------------- | | Frontend | react-best-practices, tailwind-patterns, frontend-design | | Backend | api-patterns, nodejs-best-practices, python-patterns | | Database | database-design | | Testing | testing-patterns, tdd-workflow, lint-and-validate | | DevOps | bash-linux, powershell-windows | | Architecture | clean-code, mcp-builder | | Security | (loaded by security-auditor and penetration-tester agents) | | SEO | (loaded by seo-specialist agent) |

You can edit any SKILL.md to add your team's own conventions — every agent that loads that skill will follow them automatically.

CLI Commands

npx create-raffles-it           # Install kit into current directory
npx create-raffles-it init      # Same as above
npx create-raffles-it list      # List all available agents
npx create-raffles-it help      # Show help

Activate in Claude Code

Project-local (recommended for teams):

# After running npx create-raffles-it, agents/ is detected automatically
claude .

Global (available across all projects):

# macOS / Linux
cp -r agents/ ~/.claude/agents/

# Windows (PowerShell)
Copy-Item -Recurse agents\ $env:USERPROFILE\.claude\agents\

Important: .gitignore note

If you use Cursor or Windsurf, do not add agents/ to .gitignore — the IDE won't index the agent definitions and slash commands won't appear in the chat dropdown.

To keep agents local without tracking them in Git, add to .git/info/exclude instead:

agents/
skills/
workflows/

Documentation

• Official Docs — Full documentation, guides, agent & skill reference
• Getting Started Guide — Installation, activation, and real-world examples
• npm Package — Package page on npm
• GitHub Repository — Source code, issues, contributions

Support This Project

License

MIT © trieubh