Open-source observability, built for the CLI

Debug production in your terminal.

Traceloop streams distributed traces, structured logs, and live metrics straight to your shell. No dashboard to babysit — grep a production incident like it's a local logfile.

Start free — no card
12.4M spans / day 3,400+ teams 99.99% ingest uptime 40+ auto-instrumented libraries
Observability, unbundled

Everything you need. Nothing you'll babysit.

Three pillars, one binary. Traceloop derives traces, logs, and metrics from a single OpenTelemetry stream — then hands them back to you where you already live: the command line.

Distributed traces

Follow one request across every service, queue and database hop. Waterfalls and flamegraphs render right in your terminal — no context-switch to a dashboard.

  • Span-level flamegraphs, ASCII or SVG
  • Automatic critical-path highlighting
OpenTelemetry-native

Structured logs

Every log line is stitched to its span and trace id. Filter by service, level or trace with a query syntax that feels like the grep you already know.

  • level:error svc:payments
  • Live tail, or replay any window
grep-compatible

Metrics & SLOs

RED metrics are derived automatically from your traces — rate, errors, duration. Wire up burn-rate alerts to Slack or PagerDuty without touching a single graph.

  • p50 / p95 / p99 out of the box
  • Multi-window, multi-burn-rate alerts
Prometheus export
Zero to traced

Up and streaming in 60 seconds.

Install the CLI, drop one line into your app's entrypoint, and Traceloop auto-instruments the rest. No agents, no sidecars, no sampling config to get wrong.

npm pnpm bun
# install the CLI
$ npm install -g traceloop
$ traceloop init
# install the CLI
$ pnpm add -g traceloop
$ traceloop init
# install the CLI
$ bun add -g traceloop
$ traceloop init
// traceloop.config.ts
import { defineConfig } from "traceloop";

export default defineConfig({
  service: "checkout-svc",
  env: "production",
  sampleRate: 1.0,
  instrument: "auto", // 40+ libs
});
1

Install the CLI

One global binary — macOS, Linux, Windows, or a static musl build for your CI image.

2

Import once

Add traceloop/register to your entrypoint. Auto-instrumentation wraps HTTP, DB drivers, queues and gRPC.

3

Follow production

Run traceloop up --follow and your live trace stream lands in the terminal you already have open.

Own your data

Self-host the collector or use ours. Traces never leave your VPC unless you say so — export to any OTLP backend.

Instrument anything

One import. Forty-plus libraries traced.

Traceloop speaks OpenTelemetry, so it plugs into the stack you already run — no vendor lock-in, no rewrite.

Next.js
Node.js
Python
Django
FastAPI
Flask
Express
Rails
Go
Rust
Java
.NET
Kafka
PostgreSQL
Redis
gRPC
Temporal
OpenTelemetry
Pricing

Priced per span, not per seat panic.

Start free forever. Scale when your traffic does. Every plan includes the full CLI, traces, logs and metrics.

Hobby
$0/ forever
For side projects and getting a feel for production tracing.
Start free
  • 1 project
  • 50k spans / month
  • 7-day retention
  • Community Discord support
Most popular
Team
$29/ dev · month
For product teams shipping to real users, on call for real incidents.
Start 14-day trial
  • Unlimited projects
  • 10M spans / month included
  • 30-day retention
  • Burn-rate alerts, SSO, audit log
Enterprise
Custom
For orgs with compliance, scale and on-prem requirements.
Talk to sales
  • Unlimited retention & spans
  • Self-hosted / on-prem collector
  • SAML, SCIM, 99.99% SLA
  • Dedicated Slack channel & SA
Docs

The docs read like a good README.

Copy-paste-able from the first line. Real snippets, real defaults, a quickstart you can finish before your coffee. Bring your own OTLP backend or ours.

# quickstart, in full
$ npm i -g traceloop
$ traceloop init --framework next
wrote traceloop.config.ts
patched next.config.js
$ traceloop up --follow
live — tracing 12 services