Supernova

Agent Skills for AI-Powered Development - Ship Production Apps 10x Faster.

Documentation | Security | Install Guide

---

What is Supernova?

Supernova is a collection of 27 agent skills for AI coding tools like Claude Code, Cursor, Codex, OpenCode, and Antigravity.

The Problem: Vibe-coders and non-technical builders use AI to describe ideas and generate code - but skip the foundations that real dev teams take for granted: system architecture, database migrations, security patterns, testing strategies, and production-grade integrations.

The Solution: Supernova fills these gaps with domain-specialized SOPs that guide AI agents to build real applications - not prototypes. Every skill contains production patterns, not generic templates.

---

Skills Catalog (27 Skills)

Foundation (Process & Orchestration)

Skill	What It Does
plan	Agile sprint planning, roadmap creation, ticket breakdown
orchestrator	Analyzes scope, detects complexity mode, routes to correct workflow
executor	Executes implementation tasks with built-in verification gates
parallel	Coordinates multi-agent parallel execution for complex tasks

Backend Stack

Skill	What It Does
backend	Python / FastAPI application architecture, dependency injection, middleware
api	REST (FastAPI) vs GraphQL API design based on complexity
db	PostgreSQL schema design, indexing, query optimization, RxDB for offline-first

Frontend Stack

Skill	What It Does
frontend	Next.js 14 / TypeScript / Tailwind / Shadcn/ui component architecture
ui-ux	Design systems, responsive layouts, accessibility, micro-interactions

Infrastructure & Security

Skill	What It Does
system-architecture	System design, ADRs, data modeling, API contracts
security	JWT auth, RBAC, input validation, OWASP patterns, secret scanning
devops	Docker, CI/CD pipelines, GitHub Actions, deployment workflows
infra	Terraform, Kubernetes, cloud resource provisioning

Operations

Skill	What It Does
audit	Codebase health audits, dependency review, technical debt analysis
report	Engineering reports, sprint summaries, stakeholder updates
docs	Technical documentation, API docs, user guides, changelogs

Quality & Logic

Skill	What It Does
testing	Full test pyramid — unit (pytest/Vitest), integration, E2E (Playwright)
business-logic	Domain modeling, rule engines, state machines, validation layers
state-management	Frontend (Zustand, React Query, nuqs) + Backend (Redis, sessions)

Integrations

Skill	What It Does
payments	Stripe checkout, subscriptions, webhooks with idempotency, refunds
auth-provider	Clerk, NextAuth.js OAuth, Supabase Auth — third-party auth integration
migrations	Alembic migration workflow, 3-step NOT NULL pattern, rollback SOP
file-storage	S3 / Cloudflare R2 uploads, pre-signed URLs, MIME validation, CDN
email	Resend transactional email, Jinja2 templates, Celery async sending
monitoring	Sentry error tracking, structlog, Prometheus metrics, health checks
ai-integration	LLM APIs (Claude/GPT), streaming SSE, tool use, RAG with pgvector
onboarding	Day-0 project scaffold — monorepo structure, Docker Compose, first commit

---

Compatibility Matrix

Platform	Support	Install Guide	Config
Claude Code	Full	.claude-plugin/	plugin.json
Cursor	Native	.cursor-plugin/	plugin.json
Antigravity	Full	.antigravity/	SKILL.md
Codex	Managed	.codex/	config.toml
OpenCode	Flexible	.opencode/	.opencode.json

---

Quick Start

git clone https://github.com/mrsknetwork/supernova.git
# Follow the platform-specific INSTALL.md in the dot-directories above

---

Default Tech Stack

When starting a new project, Supernova asks for confirmation before applying defaults:

Layer	Default	Alternatives
Backend	Python 3.12 + FastAPI	Django, Express
Frontend	Next.js 14 + TypeScript + Tailwind + Shadcn/ui	Vite + React
Database	PostgreSQL + SQLAlchemy 2.0	RxDB (offline-first)
API	REST (FastAPI) / GraphQL (by complexity)	tRPC
Auth	Clerk (managed) / Custom JWT	NextAuth.js, Supabase Auth
Payments	Stripe	—
CI/CD	GitHub Actions	GitLab CI

Skills always ask: "Can I use the standard Supernova stack, or do you have a preference?" before applying defaults.

---

What's Included

Directory	Purpose
skills/	27 agent skills with SOPs, evals, and references
commands/	/nova unified command entry point
hooks/	Session-start routing + security scanning hooks
.supernova/	Runtime config (modes, security, optimization)
tests/	Test suite for skill validation

---

Governance & Contributing

• Contributing Guidelines - Naming conventions and PR workflows
• Security Policy - Reporting vulnerabilities
• License - MIT Licensed

---

Philosophy

1. Production-First: Every skill targets production deployment.
2. Domain-Specialized: SOPs are specific to each technology, not generic AI templates.
3. Ask Before Assuming: Default stacks require user confirmation.
4. Evidence Over Claims: Verify results before declaring success.

---

Copyright (c) 2026 Kamesh. MIT License.