engineer-grade-agent-skills

Use when working with Apache Pulsar — designing topic hierarchies, configuring subscriptions, implementing geo-replication, deploying BookKeeper, writing Pulsar Functions, or deciding between Pulsar and Kafka

Expert Iggy message streaming — architecture, Rust SDK producers/consumers, consumer groups, offset management, server config, CLI, and production tuning. Trigger on Iggy setup questions, Rust streaming code, Iggy vs Kafka decisions, partition/offset strategy, or deployment config.

Use when working with MQTT for IoT, mobile, or real-time applications — topic design, QoS levels, broker selection, device shadows, security, or MQTT 5.0 features

Use when working with NATS messaging or JetStream persistence — pub/sub, streams, consumers, KV store, object store, clustering, or leaf nodes

Use when working with Redpanda — setting up clusters, configuring topics, tuning performance, migrating from Kafka, or integrating Redpanda Connect pipelines

Expert RobustMQ skill — multi-protocol cloud-native message queue (MQTT, Kafka-compat, AMQP) built in Rust. Trigger on RobustMQ setup, MQTT IoT streaming, Kafka-compatible migration, AMQP/RabbitMQ replacement, multi-protocol broker architecture, or RobustMQ vs alternatives decisions.

Implement Google's Agent-to-Agent (A2A) protocol for agent interoperability — invoke when the user is building multi-agent systems, agent registries, agent delegation, or integrating A2A with Spring AI or Python frameworks.

Web accessibility (a11y) — invoked when building UI components, reviewing code for WCAG compliance, fixing keyboard navigation, ARIA usage, color contrast, or screen reader support.

Production AI safety patterns for LLM applications — invoke when the user is building guardrails, defending against prompt injection, handling sensitive data, auditing tool execution, or red-teaming an LLM system.

Invoke when designing API gateways, BFF (Backend for Frontend) layers, rate limiting strategies, authentication at the edge, request aggregation, or selecting and configuring gateway products like Kong, AWS API Gateway, Traefik, or Spring Cloud Gateway.

Invoke when designing, reviewing, or refactoring services to be cloud-native — covering 12-Factor principles, stateless design, graceful shutdown, container best practices, GitOps, and cloud-native patterns like sidecar and bulkhead.

Give or receive high-quality code reviews — use when someone asks how to review a PR, write review comments, respond to reviewer feedback, structure a PR, or improve code review culture.

Expert concurrency patterns — invoke when designing or debugging multi-threaded code, async systems, actor models, channels, race conditions, deadlocks, or parallel data processing in Java, Kotlin, Python, Go, or Rust.

Expert CSS guidance — invoked when writing or reviewing CSS/SCSS for layout, responsive design, animations, modern CSS features, or performance.

Debug production incidents and hard-to-reproduce bugs — use when someone asks how to diagnose a JVM memory leak, Kubernetes crash, slow query, production exception, or needs to read thread dumps, heap dumps, or flamegraphs.

Design system architecture — invoked when building or maintaining component libraries, design tokens, theming, documentation, or distribution infrastructure.

Deep reference for distributed systems design — invoke when reasoning about consistency models, CAP tradeoffs, consensus algorithms, replication strategies, distributed transactions, or failure detection in any architecture discussion or implementation.

Implement, evaluate, or troubleshoot eBPF-based observability — including Pixie, Cilium/Hubble, Pyroscope continuous profiling, bpftrace, and BCC tools.

Expert financial systems design patterns — invoke when building payment processing, ledgers, accounting, money arithmetic, reconciliation, fraud detection, or any system where correctness of monetary values is critical.

Expert guidance for LLM fine-tuning decisions, PEFT methods (LoRA/QLoRA), dataset preparation, training, evaluation, and deployment — invoke when the user asks about training, adapting, or customizing a language model.

Frontend architecture patterns — invoked when choosing rendering strategies, structuring large applications, designing API layers, state architecture, build tooling, or testing strategy.

Design, write, test, or review Helm charts for Kubernetes — including values design, template best practices, subcharts, OCI registry, and CI testing.

Expert Kubernetes production operations — invoke when designing workloads, configuring autoscaling, setting up availability guarantees, hardening security posture, or diagnosing production issues beyond basic kubectl usage.

Set up, configure, or troubleshoot local development environments using Docker Compose, Dev Containers, Tilt, Skaffold, task runners, and environment variable management.

Invoke when selecting a NoSQL database, designing schemas for MongoDB/DynamoDB/Cassandra, modeling access patterns, or evaluating trade-offs between document, key-value, wide-column, graph, and time-series stores.

Platform engineering and Internal Developer Platform (IDP) design — invoke when designing developer platforms, golden paths, self-service infrastructure, Backstage catalogs, paved roads, policy as code, or measuring developer experience.

Expert React 18+ patterns — invoked when designing components, hooks, state management, data fetching, or performance optimization in React applications.

Expert Redis skill — invoke when designing caching strategies, selecting data structures, implementing distributed locks, reasoning about eviction policies, building Pub/Sub or Streams pipelines, or tuning Redis for production performance.

Design, implement, or review SaaS architecture patterns including multi-tenancy models, tenant isolation, billing, feature access control, and operational concerns.

Design, implement, or debug distributed sagas for multi-service transactions — use when someone asks about distributed transactions, compensating actions, saga patterns, or replacing 2PC.

Invoke when designing, configuring, or troubleshooting service-to-service communication with Istio or Linkerd — including mTLS, traffic management, canary deployments, circuit breaking, and mesh observability.

Set up, configure, or improve static analysis and code quality automation — use when someone asks about linters, SAST, SonarQube, SpotBugs, ruff, mypy, clippy, quality gates, or CI code quality checks.

Apply WebAssembly patterns for browser compute offload, WASI server-side plugins, edge functions (Cloudflare Workers, Spin), and Rust-to-Wasm compilation — including when to use Wasm and when to avoid it.

Web performance optimization — invoked when diagnosing slow pages, improving Core Web Vitals, optimizing bundles, images, caching, or server rendering strategy.

Design and implement real-time communication — choose between WebSocket, SSE, and long polling; scale with Redis pub/sub; handle presence, auth, and horizontal scaling for chat, live feeds, collaborative editing, and streaming AI output.

Create C4 Model architecture diagrams (Context, Container, Component, Code) using PlantUML, Mermaid, or Structurizr — invoked when documenting system architecture at any zoom level

Design and implement CI/CD pipelines with GitHub Actions or GitLab CI, including caching, secrets, deployment strategies, and rollback patterns

Invoked when the user needs to implement, debug, or architect consumer-driven contract testing with Pact between microservices — including broker setup, CI integration, and provider verification.

Expert database design guidance — schema design, normalization, indexing strategy, zero-downtime migrations, transactions, and query optimization for production PostgreSQL systems.

Invoked when the user needs to implement, architect, or clean up feature flags — including progressive rollout, A/B testing, kill switches, OpenFeature integration, Spring/Python SDKs, and flag lifecycle management.