Getting Started

Peekr traces your LLM calls with two lines of Python. Call peekr.instrument() once and every OpenAI, Anthropic, Gemini, and Bedrock call is captured automatically — no wrappers, no proxy.

Install

Python 3.9+ required.

pip install peekr

Or pull in an LLM client at the same time:

pip install "peekr[openai]"      # OpenAI + Peekr
pip install "peekr[anthropic]"   # Anthropic + Peekr
pip install "peekr[all]"         # every supported client

Quickstart (script)

Add two lines at the top of your entrypoint, then run your code as usual.

import peekr
peekr.instrument()

import openai
openai.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello"}],
)

Spans are written to traces.jsonl in the current directory. View them from the CLI (add --io to see inputs and outputs):

peekr view traces.jsonl
peekr view traces.jsonl --io

Quickstart (FastAPI)

Same idea for a web app — instrument first, then add the middleware to capture request-level context on every span.

import peekr
peekr.instrument()

from fastapi import FastAPI

app = FastAPI()
app.add_middleware(peekr.FastAPIMiddleware)

Connect to Peekr Cloud

Point Peekr at the hosted dashboard by passing an HTTPExporter. Spans are batched and sent in the background, so nothing blocks your app.

import peekr

peekr.instrument(
    tenant_id="acme",
    exporter=peekr.HTTPExporter(
        endpoint="https://peekr.starkspherelabs.com",
        api_key="pk_live_…",
    ),
)

import { instrument } from "@peekr/sdk";

instrument({
  exporter: {
    type: "http",
    endpoint: "https://peekr.starkspherelabs.com",
    apiKey: "pk_live_…",
  },
});

Next steps