Skill UI

Provides comprehensive patterns for creating production-grade client wrappers when integrating with OpenRouter via the OpenAI SDK. This skill covers handling advanced topics such as robust retry mechanisms, structured typing, caching strategies, and middleware implementation. It is essential for developers building stable, reliable SDKs or commercial client libraries that require enterprise-level API robustness.

This comprehensive guide provides developers with step-by-step instructions for migrating applications from direct calls to proprietary Large Language Model (LLM) APIs (such as OpenAI or Anthropic) to utilizing the unified OpenRouter gateway. It details essential code changes, including updating base URLs, managing API keys, adding custom headers, and correctly prefixing model identifiers. This ensures a seamless transition, enabling multi-model compatibility and simplified API management.

This skill automates the entire software development process overnight. By leveraging mandatory Git hooks, it strictly enforces Test-Driven Development (TDD). Use this when complex features need building, large codebases need refactoring, or bugs need fixing autonomously, guaranteeing that every commit passes rigorous testing cycles and meets high production-ready standards.

This guide details how to implement Role-Based Access Control (RBAC) for Perplexity API in enterprise environments. Since Perplexity lacks native RBAC, the solution involves building a middleware gateway. This gateway enforces granular policies, including team-specific API key usage, model restriction (e.g., limiting access to 'sonar' only), budget caps, rate limiting, and mandatory search domain filtering for compliance and security.

This guide details the recommended pattern for managing per-project plugin settings and state within a development environment. Plugins store their configuration in a dedicated `.claude/plugin-name.local.md` file. This file utilizes YAML frontmatter for structured settings (like enabling flags, retry counts, and parameters) and a Markdown body for additional context or instructions. It provides techniques for reading these settings using Bash hooks, commands, and agents.

This guide provides comprehensive strategies for implementing privacy-safe analytics using PostHog. It addresses critical compliance requirements like GDPR and CCPA, covering explicit consent management (opt-in/opt-out), real-time property sanitization (stripping PII like IP or tokens), and methods for handling Data Subject Access Requests (SAR) and the Right to Erasure. Use this to build highly compliant and privacy-preserving analytics pipelines.

An expert assistant providing deep knowledge and best practices for Prisma ORM. It covers the entire lifecycle of database interaction, including optimal schema design, relation modeling, safe migration workflows, and advanced query optimization techniques (addressing N+1 issues and implementing raw queries) across PostgreSQL, MySQL, and SQLite.

A comprehensive guide for profiling, analyzing, and optimizing Python code. It addresses identifying performance bottlenecks related to CPU usage, excessive memory consumption, I/O operations, and slow data processing. Use this skill to improve application speed and efficiency in production environments.

Provides a comprehensive pytest-focused testing strategy for Python applications, covering TDD cycles, fixtures, parametrization, coverage tools, and test infrastructure setup for reliable suites.

This comprehensive guide provides robust strategies for testing Python code using industry best practices. It covers essential techniques like pytest, fixtures, mocking, parameterization, and implementing Test-Driven Development (TDD). Use this skill when writing unit tests, setting up complex test suites, performing integration testing, or establishing continuous testing infrastructure in CI/CD pipelines.

Guides adding PyTorch FSDP2’s fully_shard integration to training loops, covering correct distributed init, bottom-up sharding, mixed precision/offload policies, DTensor-aware optimizers, and distributed checkpointing so oversized models or DeviceMesh sharding goals work reliably.

High-level PyTorch Lightning streamlines training loops with automatic distributed support (DDP/FSDP/DeepSpeed), callbacks, checkpointing, logging, and device management so you can scale from a laptop to multi-node clusters without boilerplate.