Skill UI

Phoenix Evals lets teams build evaluators for AI/LLM apps by wiring deterministic code with LLM nuance, running Python or TypeScript workloads, validating against humans, and iterating through error analysis, RAG, and production workflows.

Guides hooking Phoenix AI services with OpenInference tracing, covering setup, spans, instrumentation, and production deployment for LLM workflows using Python or TypeScript.

Acontext serves as a robust memory layer for production-grade AI agents. It provides persistent sessions, disk-based storage, and sophisticated learning spaces that distill complex conversations into reusable, structured skills. Users can manage the entire system—from project setup and login to skill uploading—via a powerful Command Line Interface (CLI) and API, significantly enhancing agents' contextual understanding and long-term knowledge retention.

A logging wrapper designed to instrument and monitor calls to the Claude API. It automatically tracks critical metrics including token usage, API latency, estimated cost, and error details. Essential for implementing robust observability patterns and setting up proactive alerts in production environments.

This pattern library provides production-ready, structured patterns for integrating with the Abridge clinical AI API. It establishes type-safe client wrappers, implements mandatory HIPAA-compliant audit logging (ensuring no Protected Health Information, or PHI, is logged), and incorporates robust error handling with exponential backoff retries. Use this when building secure, scalable, and standards-compliant healthcare AI applications.

This tool generates production-grade, spec-compliant agent definition files (.md) adhering to the Anthropic 2026 schema. It supports both the creation of new custom subagents and the validation of existing agent files against marketplace compliance rules. Use it when building complex agent architectures, reviewing agent quality, or ensuring agents meet high industry standards.

A comprehensive checklist for deploying Claude API integrations to a production environment. It covers essential best practices for reliability, observability, and cost control, including authentication procedures, advanced error handling, rate limit management, health check implementation, and infrastructure deployment strategies (e.g., canary releases) to ensure a smooth and stable launch.

Implements robust rate limiting, backoff logic, and quota management for the Anthropic Claude API. It handles 429 errors and optimizes request throughput by managing Requests Per Minute (RPM), Input Tokens Per Minute (ITPM), and Output Tokens Per Minute (OTPM). This skill provides advanced strategies including SDK retries, custom header awareness, queue-based throttling, and batch processing for stable, high-throughput production applications.

This skill provides comprehensive guides for deploying stateless, serverless API wrappers powered by Claude (Anthropic). It details how to integrate and stream responses from the Messages API across major cloud environments, including Vercel Edge Functions, Fly.io, and Google Cloud Run. It is essential for productionizing AI applications that require robust secret management, scalable infrastructure, and real-time streaming capabilities.

A comprehensive checklist designed to ensure Claude-powered applications are robust, secure, and production-ready before going live. It guides developers through critical areas, including robust error handling, API security best practices, cost control mechanisms, real-time monitoring setup, and performance optimization, ensuring a stable user experience.

A comprehensive guide to production-ready patterns for integrating with Anthropic's Claude API using both TypeScript and Python SDKs. Covers robust client setup (retries, timeouts), advanced error handling, streaming implementations, prompt caching, and batch processing to ensure high performance and cost efficiency in real-world AI applications.

Provides strategies to optimize costs when utilizing the Cohere API. This guide covers model tier selection (economy, standard, premium), implementing token budgets, reducing embedding storage costs (e.g., using int8), and establishing comprehensive usage trackers. It is essential for maintaining cost predictability in production AI deployments.