Skill UI

This skill implements a sophisticated, config-driven rules engine for OpenRouter API calls. It dynamically determines the optimal large language model (LLM) by evaluating a comprehensive context, including user subscription tier, remaining cost budget, specific task type (code, chat), required tools, and latency requirements. This is essential for building production-grade, cost-aware, and robust LLM orchestration systems.

This skill provides comprehensive tracking and analysis of OpenRouter API usage metrics. It collects critical data points such as per-request total cost, prompt and completion token counts, latency, and usage patterns. Use this skill when building financial dashboards, optimizing AI spending, monitoring API consumption, or generating detailed usage reports.

Flash Attention optimizer delivers 2-4x speedup and 10-20x memory savings for transformer attention, ideal when training or running long-context models that hit GPU memory or latency limits; supports PyTorch native scaled_dot_product_attention, the flash-attn library, H100 FP8 precision, and sliding-window attention.

This skill simulates an expert performance engineer capable of diagnosing complex bottlenecks across the entire technology stack. It covers modern observability techniques (OpenTelemetry, APM), advanced profiling (CPU, Memory, I/O), and sophisticated load testing strategies (k6, Chaos Engineering). Use this when optimizing scalability, reducing latency, improving throughput, or architecting highly reliable, performant distributed systems in cloud environments.

This skill provides comprehensive, multi-layered diagnostic scripts for resolving complex and intermittent issues with the Perplexity Sonar API. It allows users to test connectivity (DNS, TLS, TCP), profile API latency, validate citation stability, and verify search recency filters. It is crucial when standard troubleshooting methods prove insufficient for diagnosing hard-to-replicate API bugs or performance regressions.

A comprehensive guide to implementing Perplexity's API integration at various scales. It provides three validated architectural blueprints—Direct Widget, Cached Research Layer, and Multi-Query Pipeline—allowing developers to choose the optimal structure based on required query volume, latency goals, and complexity. Learn how to scale from simple search widgets to automated, high-volume research pipelines.

This skill provides comprehensive guidance on load testing and capacity planning for the Perplexity Sonar API. Users learn how to simulate high traffic volumes, respecting strict rate limits (like 50 RPM), benchmark latency, and estimate operational costs. It is essential for performance engineering, capacity architecture design, or stress testing APIs at scale.

Provides a comprehensive monitoring solution for the Perplexity Sonar API. It tracks critical performance metrics, including latency (p95), cost per query, error rates, and citation quality. This skill is essential for implementing production monitoring dashboards, configuring proactive alerts (e.g., high latency, zero citations), and ensuring the API's stability and high-quality output in real-world applications.

This guide provides advanced architectural patterns to optimize interactions with the Perplexity Sonar API. By implementing smart model routing, query hashing with time-to-live (TTL) caching, streaming responses, and rate-limited parallel processing, developers can significantly reduce latency, improve overall throughput, and manage API costs. Ideal for integrating complex, research-intensive search-augmented applications.

Fully managed, auto-scaling vector database tailored for production AI workloads such as RAG, recommendations, and semantic search, offering hybrid dense+sparse search, namespaces, metadata filters, and <100ms latency on serverless infrastructure.

Monitor the operational health of PostHog integrations by tracking critical metrics. This skill checks event ingestion rates, feature flag evaluation latency, and monitors billing volume consumption. It also provides rules for Prometheus/Grafana alerting, ensuring data pipelines remain stable and alerting on potential performance bottlenecks or exceeding API limits.

This comprehensive guide outlines advanced strategies for optimizing PostHog in production environments. Key techniques covered include implementing local feature flag evaluation to eliminate network latency, optimizing event batching for network efficiency, utilizing strategic event sampling, and writing efficient HogQL queries. Master these methods to dramatically improve data collection reliability and query speed.