Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.inference.net/llms.txt

Use this file to discover all available pages before exploring further.

This page collects the most common tracing patterns in one place. Use the integration-specific pages when you want the full setup notes for a provider or framework.

​
Direct SDK Calls

PatternGuide
OpenAI chat, tool calling, structured outputs, Responses APIOpenAI traces
Anthropic messages, tool use, prompt cachingAnthropic traces
Vercel AI SDK generateText, streamText, ToolLoopAgent, tools, structured outputVercel AI SDK traces
Stable agent.id, agent.name, and agent.role attributesAgent identity

​
Frameworks And Agents

PatternGuide
LangChain agents and toolsLangChain traces
LangGraph state graphsLangGraph traces
LangSmith traceable functionsLangSmith traces
Pydantic AI structured agentsPydantic AI traces
OpenAI Agents tools and handoffsOpenAI Agents traces
LiveKit Agents sessions, model calls, and function toolsLiveKit Agents traces
ElevenLabs conversation sessions and client toolsElevenLabs Agents traces
Cursor SDK agent runs, streamed events, and tool callsCursor SDK traces
Claude Agent SDK query loopsClaude Agent SDK traces
Claude Code CLI and SDK-style invocationsClaude Code SDK traces

​
Custom Work

Use manual spans when a library cannot be patched or when you want a product operation to group multiple automatic child spans.
TypeScript
await agentSpan(
  tracing.tracer,
  {
    agentId: "support-agent",
    name: "SupportAgent",
    role: "support",
    system: "openai",
  },
  async (span) => {
    span.setInput(userMessage);
    const result = await runAgent(userMessage);
    span.setOutput(result.answer);
  },
);
  1. Traces Quickstart
  2. OpenAI traces or Anthropic traces
  3. Manual spans
  4. Agent identity
  5. Troubleshooting