Production-grade skills library for Claude Code and Codex covering software architecture, web apps, mobile apps, SaaS, ERP, AI systems, security, UX, databases, and SDLC documentation.
The repository is designed to help Claude Code and Codex produce systems that are secure, scalable, maintainable, performance-conscious, and user-centered. It is not just a bag of examples. It is a layered engineering system.
The baseline is now aimed at world-class software development, engineering, and management: software that solves expensive real problems, feels excellent to use, operates reliably, evolves safely, and becomes more valuable as the team scales.
Skills are markdown instruction packages that encode reusable engineering judgment:
- when to use a pattern
- how to execute it
- what to avoid
- what quality gates must be met before calling the result production-ready
Active skills live under skills/<skill-name>/SKILL.md. The repository root is reserved for project documentation and operational folders such as docs/, skills/, and projects/ where relevant.
For serious implementation work, start with the repository baseline before loading stack-specific skills:
world-class-engineeringskill-composition-standardssystem-architecture-designdatabase-design-engineeringsaas-erp-system-designgit-collaboration-workflow
These skills establish the shared bar for architecture, data modeling, workflow design, release discipline, security, performance, UX, testing, operability, and engineering management. skill-composition-standards is the enforcement spine that codifies the house-style template and declares the input/output artifact contracts every other skill consumes and produces.
The baseline is intended to be executable, not aspirational: small-batch integration, risk-based testing, build-once artifact promotion, diagnosis-first telemetry, and recovery-aware release design are part of the default standard.
For high-stakes engineering work, load in this order:
world-class-engineering
-> architecture or data baseline skill
-> platform or framework skill
-> security, UX, performance, and validation companion skills
-> reliability, distributed-systems, and management skills when complexity justifies them
Typical baseline choices:
system-architecture-designfor module boundaries, contracts, ADRs, failure designdatabase-design-engineeringfor schema shape, tenancy, indexing, retention, migration safetysaas-erp-system-designfor configurable business workflows, approvals, controls, auditabilitygit-collaboration-workflowfor branch, review, merge, and release discipline
| Skill | Purpose |
|---|---|
world-class-engineering |
Shared production-grade engineering bar and release gates |
skill-composition-standards |
House-style template, cross-skill I/O contracts, and 14 canonical artifact templates |
validation-contract |
Seven evidence categories every specialist skill declares against and the canonical Release Evidence Bundle for ship readiness |
capability-matrix |
Per-domain Foundation → Implementation → Validation → Companions lookup table; load when starting any project or feature to pick the right skill stack |
system-architecture-design |
Decomposition, contracts, ADRs, failure design, scaling tradeoffs |
database-design-engineering |
Cross-engine data architecture, tenancy, indexing, migration safety, retention |
saas-erp-system-design |
Configurable SaaS and ERP workflow design, controls, auditability, extensions |
git-collaboration-workflow |
Branch, commit, PR, merge, conflict, and release discipline |
observability-monitoring |
Logs, metrics, traces, alerts, SLOs, and diagnosis-first production telemetry |
reliability-engineering |
Fault-tolerant workflow design, degradation, incident readiness, and recovery discipline |
advanced-testing-strategy |
Risk-based testing across unit, integration, contract, E2E, and release evidence |
deployment-release-engineering |
Rollout, rollback, migration-safe releases, and post-deploy verification |
distributed-systems-patterns |
Consistency, messaging, idempotency, sagas, and service-boundary tradeoffs |
engineering-management-system |
Team operating rhythm, prioritization, delegation, coaching, and delivery-system health |
| Skill | Purpose |
|---|---|
multi-tenant-saas-architecture |
Tenant isolation, audit trails, SaaS backend patterns |
modular-saas-architecture |
Pluggable business modules and per-tenant enablement |
dual-auth-rbac |
Session and JWT authentication with role-based access control |
api-design-first |
OpenAPI-first REST design, versioning, auth, caching, rate limiting |
api-error-handling |
Standardized API errors and response formatting |
saas-accounting-system |
Double-entry accounting engine patterns |
inventory-management |
Inventory, stock movement, valuation, and multi-location patterns |
| Skill | Purpose |
|---|---|
mysql-best-practices |
MySQL production patterns for schema, performance, security, HA |
mysql-data-modeling |
Universal entity patterns and business data models |
mysql-query-performance |
Query tuning, plans, indexes, and diagnosis |
mysql-administration |
Replication, backup, recovery, and operational safety |
database-internals |
Storage engine tradeoffs and core internals |
database-reliability |
SLOs, migrations, backup verification, incident runbooks |
postgresql-patterns |
Postgres-as-second-DB patterns — JSONB, full-text search, pgvector, RLS, PgBouncer, MySQL→PG translations |
| Area | Key Skills |
|---|---|
| Web frontend | webapp-gui-design, premium-ui-ux-design, form-ux-design, responsive-design, frontend-performance |
| Web backend | php-modern-standards, api-design-first, docker-development, php-security, nodejs-development, nextjs-app-router |
| Android | android-development, android-ui-ux-design, android-data-persistence, android-tdd, jetpack-compose-ui |
| iOS | ios-development, ios-ui-ux-design, ios-data-persistence, ios-tdd, swiftui-design |
| Cross-platform planning | mobile-saas-planning, mobile-reports, mobile-report-tables, mobile-rbac |
Use premium-ui-ux-design whenever perceived quality, beauty, dashboard clarity,
conversion, or commercial trust is part of the product outcome. Native mobile work
should pair android-development with android-ui-ux-design, and
ios-development with ios-ui-ux-design, so Android and iOS products get
platform-specific premium UX instead of generic responsive-web patterns.
Python sits alongside the PHP backend as a sidecar and worker to handle analytics, document generation, predictive ML, and data pipelines.
| Skill | Purpose |
|---|---|
python-modern-standards |
House style for all Python — uv, ruff, mypy, Pydantic v2, structlog, testing, security baseline |
python-saas-integration |
FastAPI sidecar + Redis worker integration with PHP, deployment on Debian, PHP↔Python contract |
python-data-analytics |
pandas/Polars analytics, cohort/funnel/retention, financial math (IRR/NPV), statistics, geospatial |
python-document-generation |
Branded Excel dashboards, Word documents, and PDF reports delivered to web, Android, and iOS |
python-ml-predictive |
Forecasting (Prophet/statsmodels), classification/regression (sklearn, XGBoost), anomaly detection, serving |
python-data-pipelines |
ETL from external APIs, OCR, PDF/image processing, scheduling, DLQs, multi-tenant pipelines |
Load order for Python work: python-modern-standards + python-saas-integration first, then the domain-specific skill.
For Python sidecars, FastAPI services, workers, queue consumers, or API integrations, python-modern-standards now loads the API/container sidecar reference. Pair it with docker-development whenever the Python work must run as a repeatable service image rather than a loose script.
JavaScript work that grows beyond simple DOM code should pair javascript-modern with javascript-patterns so object boundaries, prototypes/classes, async flows, and browser modules remain maintainable. Node services and asset build pipelines should also pair with docker-development when the runtime, Compose environment, CI image, or production artifact matters.
Use docker-development whenever a PHP, Python, JavaScript, Node, API, worker, queue, database, cache, or reverse-proxy service needs a repeatable local environment or production image. It sets the baseline for deterministic dependency installs, Docker Compose orchestration, dev/prod image separation, secrets hygiene, image scanning, registry tagging, and build-once promotion.
Beyond typescript-mastery (type system) and typescript-design-patterns (GoF):
| Skill | Purpose |
|---|---|
typescript-effective |
Production idioms — strict tsconfig, effective-TS items, clean code, Zod boundaries, migration from JS, build performance, anti-patterns |
typescript-full-stack |
End-to-end TS — Fastify, tRPC, Prisma/Drizzle, Zod shared schemas, turborepo, auth, Docker |
| Skill | Purpose |
|---|---|
kubernetes-fundamentals |
Core objects, kubectl, probes, ingress, when K8s is right vs alternatives, EKS/GKE/kind |
kubernetes-production |
Helm, autoscaling, StatefulSets, external-secrets, observability, RBAC + PSS, NetworkPolicies, Velero, cost control |
kubernetes-saas-delivery |
Multi-tenancy models, namespace isolation, ArgoCD GitOps, progressive delivery, tenant onboarding + offboarding, cost allocation |
Extends existing gis-mapping (Leaflet) with:
| Skill | Purpose |
|---|---|
gis-postgis-backend |
PostGIS schema, SRID choice, spatial indexes, MVT tiles, geocoding, hybrid MySQL+PostGIS, RLS tenant isolation |
gis-maps-integration |
Google Maps JavaScript API + Mapbox GL — setup, Places, directions, styling, offline, cost control, accessibility |
gis-enterprise-domain |
ArcGIS Enterprise admin + real-estate GIS patterns (property search, neighbourhood analysis, catchment, market heatmaps) |
| Skill | Purpose |
|---|---|
saas-sales-organization |
Sales motions, roles (SDR/BDR/AE/CSM/SE), pipeline stages, territory + quota + commission design, forecasting, hiring |
saas-subscription-mastery |
Subscription business strategy — retention point, 29-step framework, activation, engagement loops, churn prevention, expansion, billing provider selection |
| Skill | Scope |
|---|---|
vibe-security-skill |
Secure coding baseline for web-connected systems |
web-app-security-audit |
Structured security review for web apps |
php-security |
PHP-specific security patterns |
ai-security |
LLM and AI integration security controls |
llm-security |
Prompt injection, trust boundaries, output validation |
skill-safety-audit |
Safety checks for skills themselves |
| Skill | Purpose |
|---|---|
laws-of-ux |
Named UX laws and design implications |
ux-writing |
Microcopy, empty states, errors, loading, confirmations |
ux-psychology |
Cognitive foundations for product design |
practical-ui-design |
Visual design systems, layout, typography, color |
interaction-design-patterns |
Reusable interaction patterns |
design-audit |
Structured UI quality review |
tabler-email-templates |
80 production HTML email templates (Tabler 3.0) — welcome, invoice, OTP, magic-link, password reset, newsletter, promotions, etc. Light + dark, mobile-responsive, cross-client tested |
| Skill | Purpose |
|---|---|
ai-economic-value-engine |
AI opportunity discovery, ROI, value scoring, and product strategy |
ai-app-architecture |
Production AI application architecture, governance, workflow-vs-agent decisions |
ai-agents-tools |
Agent platforms, MCP-style tool contracts, approvals, and auditability |
ai-evaluation |
Evaluation contracts, release gates, monitoring, and regression control |
ai-web-apps |
AI-enabled web app patterns, budgets, streaming, tool use |
ai-assisted-development |
Multi-agent development workflows |
ai-error-handling |
Validation stacks for AI output |
ai-error-prevention |
Trust-but-verify workflows |
orchestration-best-practices |
Multi-step AI workflow coordination |
ai-rag-patterns |
Retrieval-augmented generation patterns |
vector-databases |
Embedding model selection, chunking, hybrid search, reranking, engine choice (pgvector/Qdrant/Pinecone/Weaviate) |
rag-implementation |
Naive→Advanced→Modular RAG progression, query transforms, corrective RAG, RAGAS evaluation, multi-tenant isolation, cost levers |
- Web application:
world-class-engineering+system-architecture-design+api-design-first+database-design-engineering+docker-development+vibe-security-skill - SaaS or ERP platform:
world-class-engineering+saas-erp-system-design+modular-saas-architecture+multi-tenant-saas-architecture+database-design-engineering - Mobile-backed product:
world-class-engineering+system-architecture-design+android-developmentorios-development+ security and persistence skills - AI-enabled application:
world-class-engineering+ai-economic-value-engine+system-architecture-design+ai-app-architecture+ai-evaluation+ai-security+frontend-performance - Production release workflow:
world-class-engineering+advanced-testing-strategy+observability-monitoring+deployment-release-engineering - Distributed workflow or service platform:
world-class-engineering+system-architecture-design+distributed-systems-patterns+reliability-engineering - Engineering leadership and delivery improvement:
world-class-engineering+engineering-management-system+git-collaboration-workflow - Python analytics + document generation add-on to a PHP SaaS:
python-modern-standards+python-saas-integration+python-data-analytics+python-document-generation - Python predictive ML feature:
python-modern-standards+python-saas-integration+python-ml-predictive+python-data-analytics - Python ETL / OCR / data ingestion:
python-modern-standards+python-saas-integration+python-data-pipelines - PHP or Node API service:
php-modern-standardsornodejs-development+api-design-first+docker-development+advanced-testing-strategy+deployment-release-engineering
Reference skills directly in the prompt:
Use the android-development skill to review this feature.
Use world-class-engineering and system-architecture-design before proposing the backend design.
Reference the skills from your project-level CLAUDE.md or AGENTS.md:
Load these skills from ~/.claude/skills/skills:
- world-class-engineering
- system-architecture-design
- database-design-engineering
- api-design-firstEach skill should follow this shape:
---
name: skill-name
description: Use when ...
---Then include:
- scope and activation clues
- workflow or decision rules
- standards and anti-patterns
- references to deeper files in
references/ - the portable execution contract used by both Claude Code and Codex
- Claude Code can continue using the repository through
CLAUDE.mdand direct skill invocation. - Codex can use the same skill folders through root
AGENTS.mdrouting plus each skill'sSKILL.md. - Skills are stored in the standard
skills/<skill-name>/SKILL.mdlayout. - Optional plugin notes may remain in skills, but they must not block execution in Codex.
- Keep
SKILL.mdentrypoints and top-level repository guidance under 500 lines, and split deeper reference docs as they are refreshed. SKILL.mdshould contain execution logic, not textbook-length explanation.- Deep detail belongs in
references/. - Engineering skills should align with
world-class-engineering. - New skills should update the top-level repository docs when they materially change the system.
- 236 skills across architecture, web, mobile, Python analytics/ML/documents, full-stack TypeScript, Kubernetes, GIS, AI, security, UX, data, SaaS business, and SDLC
- full-stack coverage for web apps, mobile apps, SaaS, ERP, APIs, and data systems
- layered baseline for architecture, data design, delivery workflow, and production quality
Peter Bamuhigire
Built for Claude Code and Codex.