TraceLLM User App Example

It creates a support workflow trace, fetches real HTTP data, calls OpenAI, Anthropic Claude, or Gemini, records tool/retrieval/LLM spans, then sends the timeline to your local TraceLLM backend.

Run#

cd C:\dev\tracellm
$env:TRACELLM_ENDPOINT="http://localhost:4319"
$env:TRACELLM_API_KEY="trllm_your_key_from_the_ui"

OpenAI:

$env:LLM_PROVIDER="openai"
$env:OPENAI_API_KEY="sk_your_openai_key"
$env:OPENAI_MODEL="gpt-4.1-mini"
pnpm example:user-app

Anthropic Claude:

$env:LLM_PROVIDER="anthropic"
$env:ANTHROPIC_API_KEY="sk-ant-your_key"
$env:ANTHROPIC_MODEL="claude-3-5-sonnet-latest"
pnpm example:user-app

Gemini:

$env:LLM_PROVIDER="gemini"
$env:GEMINI_API_KEY="your_gemini_key"
$env:GEMINI_MODEL="gemini-1.5-flash"
pnpm example:user-app

Open the TraceLLM UI and refresh the session list. You should see Customer support answer workflow.

The LLM span should be named by provider, for example openai.chat.generate, anthropic.chat.generate, or gemini.chat.generate, and include real token usage when the provider returns usage metadata.