$ npm i routex — 12kB agent, zero config
Routex traces every call across your services, flags the 500s before your users tweet about them, and replays any request with one command.
Free tier: 100k requests/month · No credit card
12kB
Agent overhead
<1ms
Added latency per span
14 days
Traces retained free
4.2k
GitHub stars
Observability that feels like a dev tool, not an enterprise sales call.
Tail logs, replay requests and diff deploys from your terminal. The UI is optional.
Query traces by status, latency, user, or request body. Answers in milliseconds.
Routex correlates errors with the exact commit that introduced them.
Burn-rate alerts on the metrics that matter, p50 to p99.
Grouped, deduplicated, and routed to the on-call human with the trace already attached.
| Feature | Routex | Self-hosted ELK |
|---|---|---|
| Time to first trace | 5 minutes | A weekend |
| Infra to manage | None | Your problem |
| CLI-first workflow | ||
| Per-request replay | ||
| Cost at 10M spans/mo | $89 | $400+ in infra |
Connect the tools you already use in a couple of clicks.
Pricing
Start free. Pay for volume, not seats.
Side projects and staging.
$0
Production workloads.
$89 /mo
Compliance and volume.
Custom
Usage overage is billed at $6 per extra million requests. No surprises.
FAQ
The agent is 12kB and adds under 1ms per span. It batches and ships asynchronously — if it can’t keep up, it drops telemetry before it ever touches your request path.
First-party SDKs for Node, Python, Go and Rust. Any OpenTelemetry-compatible service works too — just point the OTLP exporter at us.
Enterprise plans include an on-prem collector. The SaaS store is multi-tenant with per-project encryption keys.
Replay rewrites POSTs and other non-idempotent calls to your staging target by default. Point it at production only if you enjoy danger — we make you type the URL twice.
Install the agent, ship a request, open the trace. Five minutes, no sales call.