Skill UI

This skill provides comprehensive reference architectures for building production-grade AI applications using OpenRouter as a unified LLM gateway. It covers three scaling patterns: simple single-service, standard microservice (with caching and budget enforcement), and enterprise event-driven systems (with message queues and workers). Use this when planning system design, reviewing complex AI pipelines, or ensuring scalability and cost control in your LLM integration.

A comprehensive guide and implementation examples for handling API rate limits (HTTP 429) when interacting with the Perplexity Sonar API. This resource provides multiple robust techniques, including exponential backoff with jitter, queue-based rate limiting, and custom Token Bucket implementation, ensuring reliable and high-throughput API calls for developers using TypeScript and Python.

This skill enables the development of robust, event-driven pipelines interacting with the Perplexity Sonar API. Since Perplexity lacks native webhooks, this system simulates event patterns using advanced techniques: Streaming Server-Sent Events (SSE) for real-time data, job queues (like BullMQ/SQS) for asynchronous batch research, and cron-triggered monitoring for scheduled news and trend alerts. It provides patterns for building complex, reliable, and scalable AI applications.

Automated guidance for setting up RabbitMQ queues in backend services, covering patterns, code snippets, and validation for Node.js, Python, Go, and microservices messaging scenarios.

Automates AWS SQS queue setup by guiding users through best-practice architecture, configuration, and code output, triggering whenever someone asks about queue creation or patterns.

A comprehensive guide to implementing enterprise-grade Supabase architectures. This skill covers advanced patterns like monorepo layout, multi-tenant Row Level Security (RLS) isolation, microservices integration, and operational patterns (caching, queues, audit logging). Use this when building complex, scalable, and production-ready applications on Supabase.

This pattern demonstrates how to build a robust, thread-safe data persistence layer in Swift using the 'actor' model. It combines fast in-memory caching with reliable file-backed storage, automatically eliminating data races and the need for manual locks. Ideal for offline-first applications and managing shared mutable state in modern Swift concurrency.

A comprehensive utility suite for handling external API rate limits, specifically designed for services like TwinMind. It implements advanced patterns including exponential backoff with jitter and request queue management. Use this to ensure resilient, high-throughput, and stable interactions with resource-constrained AI APIs, minimizing 429 errors.

Upstash QStash expert handles serverless message queues, scheduled HTTP jobs, retries, deduplication, and callback tracking so you can deliver asynchronous tasks without managing infrastructure.

This skill fetches unread GitHub Pull Requests (PRs) that require review from a specified team or were opened by a team member. It is ideal for users who need to check their team's review queue, find PRs needing attention, or automate the process of reviewing open code submissions without manually browsing GitHub notifications. Requires GitHub CLI authentication.

This advanced pipeline orchestrates complex feature development based on Request for Comments (RFCs). It breaks down monolithic changes into small, independently verifiable work units, managing dependencies through a Directed Acyclic Graph (DAG). It includes multiple quality gates, dedicated unit testing, and a safe merge queue to ensure robust integration and controlled deployment.

A comprehensive guide for authorized penetration testers on identifying and exploiting insecure deserialization vulnerabilities across multiple languages (Java, PHP, Python, .NET). Covers techniques for generating payloads using tools like ysoserial and phpggc, and testing various application components such as cookies, HTTP parameters, and message queues to achieve potential Remote Code Execution (RCE).