Phoenix Tracing

This is your complete reference for instrumenting LLM apps with Phoenix's OpenInference tracing. It covers both Python and TypeScript, walking you through auto-instrumentation for popular frameworks like OpenAI and LangChain, plus manual span creation when you need fine-grained control. The real value is in the nine span type specs (LLM, chain, retriever, tool, agent, etc.) with full attribute schemas, so you know exactly what metadata to attach. Production patterns for batching and PII masking are here too. If you're shipping an LLM app and need observability beyond print statements, this gives you the complete OpenInference implementation without hunting through docs.

Install to Claude Code

npx -y skills add github/awesome-copilot --skill phoenix-tracing --agent claude-code