Stackwise Skills
End-to-end development skills for Claude Code - frontend, backend, mobile, infra, and QA - one plugin install.
Every skill is framework-specific, production-focused, and follows one enforced template, so Claude loads exactly the expertise your stack needs and nothing else. Ship a new feature or fix a bug with the same pipeline across web, mobile, and infrastructure - including the step most skill packs skip: verifying the change on a real browser or real device before you call it done.
Install
/plugin marketplace add Aarvion-AI/stackwise-skills
/plugin install stackwise@stackwise-skills
Skills activate automatically based on what you're working on - open a .tsx file and ask for a feature, and react-expert loads; touch a main.tf, and terraform-expert loads. No manual invocation needed.
What's inside
| Category | Skill | Covers |
|---|---|---|
| Core | engineering-rules |
Repo-wide code rules: reusability, readability, no AI slop, dead-code removal, why-only comments |
| Frontend | react-expert |
React 19, Next.js App Router, Server Components, state management, performance |
| Frontend | vue-expert |
Vue 3.5, Composition API, Nuxt 4, Pinia |
| Backend | nestjs-expert |
NestJS 11, modules/DI, validation, TypeORM/Prisma, auth, testing |
| Backend | fastapi-expert |
FastAPI, Pydantic v2, async SQLAlchemy, dependency injection, testing |
| Backend | fastify-expert |
Fastify 5, plugins/encapsulation, TypeBox validation, hooks, testing |
| Backend | go-expert |
Go 1.23+, net/http services, errors/context, concurrency, pgx/sqlc |
| Mobile | flutter-expert |
Flutter 3.x, Riverpod/Bloc, GoRouter, platform channels |
| Mobile | react-native-expert |
React Native 0.7x, Expo Router, native modules, platform handling |
| Mobile | swiftui-expert |
iOS native: SwiftUI, Swift 6 concurrency, Observation, SwiftData |
| Mobile | jetpack-compose-expert |
Android native: Compose, state/recomposition, Hilt, Navigation |
| Infra | terraform-expert |
Terraform 1.x, module design, state management, plan review, CI |
| QA (web) | playwright-expert |
Playwright E2E, fixtures, network mocking, visual regression, CI |
| QA (mobile) | mobile-qa-expert |
Appium 2, real-device clouds (AWS Device Farm, BrowserStack), unattended flows |
Each skill is a folder: a lean SKILL.md (when to use, core workflow with verification loops, key patterns) plus references/ files that load only when the task needs them - so 9 skills today or 50 tomorrow cost the same context.
Why this exists
Most skill packs cover writing code. Real feature work is write → wire the API → migrate the schema → change the infra → prove it works. Stackwise skills bake verification into every workflow step: type-check until clean, run the test suite, drive the actual app. The mobile QA skill runs your change on a physical device in the cloud - the difference between "compiles" and "ships."
Roadmap - contribute a framework
The skill template and CI validation make adding a framework a well-defined PR: claim it in an issue, copy templates/SKILL_TEMPLATE.md, write 3-6 references, and run the validator. Full walkthrough in CONTRIBUTING.md.
- React · Vue · NestJS · FastAPI · Fastify · Go · Flutter · React Native · SwiftUI · Jetpack Compose · Terraform · Playwright · Appium
- Frontend: Angular, Svelte/SvelteKit, Astro
- Backend: Django, Spring Boot, Rails, Laravel, .NET, Elixir/Phoenix
- Mobile: Kotlin Multiplatform, native watchOS/Wear OS, mobile release engineering (Fastlane, store rollouts)
- Infra: AWS CDK, Pulumi, Kubernetes/Helm, GitHub Actions, Docker
- QA: Cypress, Maestro, Detox, k6 (load), pytest/Jest deep-dives
- Data: PostgreSQL, Redis, Kafka, dbt
Contributing
New frameworks, version bumps on existing skills, and corrections to guidance that has gone stale are all welcome. Start here:
- Good first issues - unclaimed frameworks, scoped and ready to pick up
- CONTRIBUTING.md - the skill contract, how to run a skill from a local checkout, and the quality bar reviewers apply
- Claim a framework before you start, so two people do not write the same skill
- Report stale guidance when a framework moves and a skill does not
Scaffolding is one command, and the validator tells you what is still missing:
python3 scripts/new_skill.py <framework> --category frontend --refs a b c
python3 scripts/validate_skills.py
Every PR runs the same validator, which enforces the template contract: a "Use when" description, a verification loop in the Core Workflow, and every reference file listed in the Reference Guide table. Run it locally and your PR is already green.
Land one PR and you get push access to the repo, so your second contribution does not need a fork.
By participating you agree to the Code of Conduct. Security concerns go through SECURITY.md, not a public issue.
License
MIT. See LICENSE. Contributions are accepted under the same license.
No comments yet
Be the first to share your take.