Claude Skills by Dev-Toolbelt

Creates and validates agent files for the dev-team-agents repository. Enforces all authoring standards from CLAUDE.md — frontmatter, model assignment, required sections, line limit, and Worktree Isolation for coding agents. Use when creating a new agent or auditing an existing one for compliance.

Guides the pre-release process for the dev-team-agents repository. Runs consistency checks, validates README sync, determines the correct version bump, and creates the git tag. Use before any version tag is created.

WCAG accessibility — load only when explicitly required.

REST/GraphQL design — naming, status codes, pagination, idempotency.

API versioning — compatibility rules and deprecation lifecycle.

Where architecture documents live, what each contains, and the conformance report format.

Async jobs — idempotency, retry strategy, DLQ, observability.

Caching — in-memory, distributed, CDN, invalidation, TTL.

Component patterns — container/presentational for React/Vue/Angular.

CSS quality — specificity, magic numbers, responsive. Agnostic.

Detect and prevent duplicate/redundant API calls in SPA/SSG frontends — tool-agnostic.

Design patterns — KISS, YAGNI, DRY, SOLID, GoF, DDD reference.

Event-driven — sourcing, CQRS, saga, idempotency, schema registry.

Feature flags — types, targeting rules, rollout, lifecycle.

Form design — state, validation, error feedback. Framework-agnostic.

Base frontend code quality rules — component size, state, a11y, performance, type safety.

Frontend patterns — debounce, double-submit prevention, error boundaries.

GraphQL — schema, resolvers, N+1/DataLoader, pagination.

i18n/l10n — locale detection, translations, pluralization, RTL.

Ingestion API — inbound data boundary design, validation, backpressure, idempotency, DLQ.

LLM/RAG in product — retrieval, chunking, prompt versioning, eval, cost, failure modes.

Monorepo — workspace tooling, package boundaries, CI optimization.

Chunked file upload — pre-signed URLs, chunk management, retry.

Naming and file structure — framework-agnostic with per-framework notes.

Object Calisthenics — 9 rules for encapsulation and OOP simplicity.

SLO/SLI — error budgets, alerting, four golden signals.

Delegation mechanics — execution gate, autonomous sprint protocol, agent roster, spawn rules.

Core Web Vitals, bundle budgets, Lighthouse CI, image optimization.

Rate limiting — algorithms, tiers, backends, headers, failure modes.

Resilience — circuit breaker, retry, bulkhead, health checks.

One controller per HTTP action — naming, structure, DI, routing.

Server-Sent Events — text/event-stream format, EventSource, reconnection, keep-alive, CORS.

Frontend state — where it lives, library selection, architecture.

Cassandra — partition keys, CQL, consistency levels, CLI.

Database selection — strengths and use-case mapping by type.

Zero-downtime migrations — expand/contract and rollback.

MongoDB — schema design, indexes, aggregation, sharding, Atlas.

MySQL/MariaDB — InnoDB, indexing, replication, JSON, query tuning.

PostgreSQL — MVCC, indexes, partitioning, JSONB, query tuning.

Redis — data structures, keyspace, persistence, clustering, pub/sub.

SQLite — WAL mode, JSON1, FTS5, indexes, CLI.

SQL Server — T-SQL, indexes, Always On AG, CDC, columnstore.

Design system audit — visual patterns, consistency; Design/Consultive.

Distinctive, production-grade frontend interfaces.

Mobile-first responsive UI specs — breakpoints, navigation, spacing, touch targets.

Technical/on-page SEO, Core Web Vitals, structured data, GEO/LLM SEO, and quality-gate signals.

Web interface review — accessibility, UX, design best practices.

AWS — ECS, EC2, RDS, S3, CloudFront, ALB, IAM best practices.

Azure — Container Apps, AKS, ACR, Azure SQL, Cosmos DB, managed identity.

Shared CI/CD — stages, quality gates, secret management, rollback.