usetrace 0.1.0

Lightweight tracing SDK for LLM applications

usetrace

Lightweight tracing SDK for LLM applications. Add @tracer.observe() to your functions and get visual attribution maps that answer "Why did the LLM say that?"

Install

pip install usetrace

Quick Start

from usetrace import Trace

tracer = Trace(api_key="your-key", base_url="https://api.use-trace.com")

@tracer.observe(span_type="llm", model="gpt-4o")
def my_llm_function(prompt: str) -> str:
    return openai.chat.completions.create(...)

Every traced call is captured — inputs, outputs, latency, token usage — and sent to the Trace dashboard where you can inspect full execution trees and per-token attribution maps.

Dashboard

Sign up and explore your traces at use-trace.com.

License

Apache-2.0