Introduction

numbrs is an open-source personal metrics platform for people who self-host things and want visibility into what’s actually happening. It connects to your Nostr relays, AI usage logs, media servers, and anything you can script — and puts it all on one dashboard you control.

It’s built for individuals. Not teams. Not enterprises. Just you.

Managed vs self-hosted

You can run numbrs two ways:

Managed — Sign up at numbrs.lol. Nothing to deploy. Your data lives in Supabase with row-level security enforcing that only you can read it. Free to start.

Self-hosted — Fork satsdisco/numbrs on GitHub, create a Supabase project, deploy to Vercel. You own the database entirely. See the Self-Hosting guide for step-by-step instructions.

Both modes are identical in functionality. The only difference is who runs the infrastructure.

What you can track

Feature	What it does	Setup required
Relay monitoring	Connect latency, event latency, uptime % for any Nostr relay	UI — just paste a wss:// URL
Uptime monitoring	HTTP endpoint checks, incident history, response time trends	UI — just paste a URL
Claude AI usage	Token counts, costs per session, by project — Claude Code + OpenClaw	Collector script + cron
Jellyfin	Play history, library stats, user activity	Collector script + cron
Plex	Play history, library stats, active streams	Collector script + cron
Custom metrics	Push any number via HTTP POST, from scripts or CI	Just curl
Dashboard builder	Drag-and-drop panels, line/area/stat/gauge types, public sharing	Built-in
Alerts	Threshold-based notifications when metrics cross limits	UI — configure per metric

Who this is for

Nostr developers and relay operators — You’re running infrastructure and you need to know if your relay is up, how fast it’s responding, and whether it’s degrading over time. numbrs checks every 5 minutes and keeps the history.

Heavy Claude users — If you’re on Claude Max or running an AI assistant that burns tokens all day, it’s surprisingly hard to know where the spend is going. numbrs reads your local JSONL session files and surfaces costs by project and model, automatically.

Homelab people — You’re already running Jellyfin or Plex, you’ve got cron jobs doing things, you maybe have some servers you care about. numbrs gives you a dashboard for all of it without standing up Prometheus and Grafana.

Indie hackers who want metrics without the overhead — Push a metric from a GitHub Action, a cron script, or a Python script. See it on a dashboard. That’s it.

How data flows

Everything in numbrs follows one of two paths:

Native integrations (relay monitoring, uptime monitoring) — numbrs’ servers make the checks directly. You configure them in the UI and data appears automatically.

Collector-based integrations (Claude, Jellyfin, Plex, custom metrics) — A small script runs on your machine, reads local data, and pushes it to your Supabase database via REST API. You schedule this script with cron or launchd. The script is designed to run every 5 minutes and process only recent changes.

Your machine                    Supabase              numbrs dashboard
────────────────────            ──────────            ─────────────────
~/.claude/projects/  ──┐
~/.openclaw/sessions ──┤→ collector script → REST API → database → charts
Jellyfin/Plex API    ──┘
Your custom scripts  ────────────────────────────────────────────────────→

All data is private by default. Row-level security ensures your data is only accessible to your account.

Where to start

The fastest path: Quick Start — sign up, add an uptime monitor, push a custom metric, and build a dashboard. Takes about 5 minutes.

If you want to self-host first: Self-Hosting guide.

For AI usage tracking (the most popular integration): Claude AI Usage.