We ship TypeScript, Python, and Rust SDKs. All accept the same environment variables and produce compatible traces.

# node / deno / bun
pnpm add loupe-sdk

# python
pip install loupe

# rust
cargo add loupe

Create a project in the console; copy the project id and the API key. Set them as environment variables.

export LOUPE_API_KEY=lp_sk_live_abc123…
export LOUPE_PROJECT=support-bot
export LOUPE_ENV=prod

For self-hosted, point LOUPE_URL at your instance; no other changes.

No code changes downstream — all your chat.completions.create calls, tool calls, and retrievals now produce a trace.

import { loupe } from "loupe-sdk";
import OpenAI from "openai";

const ai = loupe(new OpenAI());

// use it exactly as before — every call is a span
const resp = await ai.chat.completions.create({
  model: "gpt-4o-mini",
  messages: [{ role: "user", content: input }],
});

Or use framework-specific integrations for LangChain, LlamaIndex, Vercel AI SDK, and Anthropic/Gemini directly.

Run one request. Open the console — a trace should land in under a second.

curl -X POST https://api.yourdomain.com/chat \
  -d '{"message": "where is my order?"}'

# in another shell
loupe tail --project support-bot

✓ trace t_01jjq… · 1.8s · 1.17k tokens · $0.0019