lumen
100-request burst engine, now public
The truth
about your
endpoints.
lumen fires 100 concurrent probes at any HTTP target, then unpacks the timeline — DNS, TCP, TLS, TTFB and percentile latency in microseconds. Built on a Go engine, orchestrated through BullMQ, persisted with Drizzle.
Run your first burst
See how it works
Trusted by infra teams that ship at scale
Boxworks
Cipher
Facet
Quadrant
Hexon
Triade
Boxworks
Cipher
Facet
Quadrant
Hexon
Triade
02
-
Observability
Microsecond truth, stitched together live.
Every probe is captured, queued, and fanned-out across workers. The result lands on a dashboard built for operators — no toy graphs, no vanity numbers.
REQUESTS PROBED
2,840,372
+12.4%
ENDPOINTS MONITORED
184
+8
MEDIAN P95 (MS)
146
−4ms
REGRESSIONS CAUGHT
38
this week
03
-
capabilities
Built like a infrastructure,
Not a side project.
Six primitives, one runtime. Every layer chosen for the worst-case on-call night.
01
A Go engine, not a wrapper.
Our CLI is purpose-built around net/http/httptrace — every connect, write, and read is timestamped at the kernel boundary.
lumen burst api.example.com \
--concurrency 100 \
--duration 30s \
--emit json
02
Async by design.
BullMQ + Upstash Redis schedule, deduplicate, and retry probes. Your dashboard never blocks on a long check.
03
Type-safe data layer.
Drizzle ORM against Postgres. Migrations stay honest, queries stay fast.
04
Polyglot monorepo.
Turborepo + pnpm workspaces. Go binary, Node workers, Next.js dashboard — one push, one pipeline.
05
P95 / P99, in your face.
Percentiles aren't a tooltip — they're the headline. Set SLOs and lumen burns when they break.
06
CI-native CLI.
Drop the binary in any pipeline. JSON, Prometheus, OTel exporters out of the box.
04
—
FLOW
Four steps,end to end.

No SDK, no instrumentation, no rewriting. lumen treats your endpoint exactly the way the rest of the internet does — and shows you what they see.

01

Drop the URL in.

No agent, no SDK, no rewriting. Just paste your API endpoint and watch it come alive. Run it anywhere laptop, CI/CD, cloud anywhere you can reach.

02

Burst 100 concurrent probes.

The engine fires a controlled burst, capturing DNS, TCP, TLS, and TTFB at the kernel-trace level. BullMQ orchestrates fan-out across regions if you scale up.

03

Persist with Drizzle + Postgres.

Every probe is hashed, normalized, and committed type-safely. Your historical data stays queryable, exportable, replayable.

04

Watch the dashboard breathe.

A high-fidelity Next.js + Framer Motion dashboard surfaces percentile latency, regressions and SLO breaches the moment they happen.

05
—
PRICING
Pay for telemetry,not for seats.

Honest, simple tiers. Cancel anytime. Open source CLI forever.

Solo

For an individual operator running checks from a laptop.

$0/mo billed annually
  • Single binary CLI
  • Up to 25 endpoints
  • 7-day metric retention
  • JSON & Prometheus export
  • Community Discord
MOST TEAMS

Team

For startups and platform teams running production traffic.

$24/mo billed annually
  • Everything in Solo
  • Unlimited endpoints
  • 90-day retention + replay
  • BullMQ region fan-out (3 regions)
  • Slack & PagerDuty alerts
  • SSO (Google + Microsoft)

Scale

For SREs at scale with audit, residency and SLO contracts.

Custom
  • Everything in Team
  • Self-hosted & dedicated cloud
  • Custom regions worldwide
  • Audit log + SOC2 II
  • Engineering-led onboarding
  • Bring-your-own Postgres + Redis
06
—
VOICES
The teams whotrusted the burst.

We replaced three internal scripts and a legacy synthetic-monitoring vendor with a single lumen burst. The TTFB delta on our checkout API was a 4-line PR.

Alina Petrova
Alina Petrova
STAFF SRE, NORTHWAVE

P99 was lying to us for months. lumen surfaced a TLS renegotiation issue on a single edge node — caught it on the first burst.

Patricia Wisewood
Patricia Wisewood
PLATFORM ENGINEER, TIDEFLOW

The dashboard reads like a piece of design software, not a Grafana clone. My on-call rotation actually opens it on Sundays.

Daniel Okafor
Daniel Okafor
VP ENGINEERING, STRATUS