modra40/claude-codex-skills-directory
Overview
This skill is the ultimate RabbitMQ expertise pack for designing, deploying, and operating production-grade message broker systems. It combines advanced messaging patterns, high-availability configurations, performance engineering, security hardening, monitoring, and troubleshooting playbooks. Use it to harden systems, debug complex failures, and scale RabbitMQ for demanding workloads.
How this skill works
I provide concrete, battle-tested patterns and configuration snippets that inspect and improve connection/channel usage, queue topology, HA replication, and client-side behavior. The skill explains what to tune and why—prefetch formulas, memory thresholds, queue types (quorum/stream), DLX and retry flows, and batch or async publishing strategies. It also covers operational checks, Prometheus alerts, and practical remediation steps for alarms, partitions, and backlogs.
When to use it
- Designing a resilient RabbitMQ architecture for production
- Optimizing throughput and latency for publishers and consumers
- Implementing reliable delivery patterns (DLX, retries, confirms)
- Hardening security: TLS/mTLS, OAuth2/LDAP, certificate rotation
- Investigating memory/disk alarms, queue backlogs, or network partitions
- Planning migrations, multi-tenant vhost design, or HA upgrades
Best practices
- Use long-lived connections and a connection pool; avoid one-connection-per-message
- Give each thread a dedicated channel; enable publisher confirms for reliability
- Prefer quorum queues for critical durable queues and streams for high-throughput replayable data
- Tune prefetch using measured RTT and processing time; batch publishes and use async confirms for throughput
- Set x-max-length/x-overflow and x-message-ttl to prevent unbounded growth; use DLX chains for retries
- Monitor key metrics (memory ratio, queue backlog, unacked, consumer utilization) and alert before thresholds are hit
Example use cases
- Implementing reliable order processing with publisher confirms, idempotency keys, and DLQ handling
- Scaling a metrics pipeline using stream queues and batching for 10x throughput
- Migrating classic queues to quorum queues with zero-downtime strategies
- Configuring delayed/scheduled messages using plugin or TTL+DLX when scheduling is required
- Diagnosing and resolving memory alarms, consumer starvation, or split-brain scenarios in a clustered deployment
FAQ
Use priority queues sparingly for small priority ranges; they add overhead. Prefer separate queues and consumer pools when you need strict isolation or large priority sets.
Are quorum queues always better than classic?
For durability and HA yes—quorum queues provide Raft replication. Classic queues may still be fine for transient workloads or when plugin compatibility (e.g., priority on classic) is required.
5 skills
This skill delivers expert RabbitMQ guidance for production-grade messaging, covering patterns, HA, performance, security, monitoring, and troubleshooting.
This skill guides Nuxt 3 and TanStack Query mastery for production-ready architecture, performance, and clean code through expert patterns.
This skill helps architects and developers apply senior Go playbooks to design, review, test, and productionize services for reliability and security.
This skill provides expert SolidJS and SolidStart guidance to build production-ready, scalable apps with fine-grained reactivity, SSR support, and
This skill acts as an elite AI/ML engineer to design production-grade ML systems, fine-tune models, and optimize pipelines end-to-end.
