agg

Simple analytics and product insights for your project.

Page views, visitors, top pages and referrers out of the box. Then count what matters for your product: feature usage, sign-ups per plan, purchases per product, bestseller per category, over sliding windows from 5 minutes to 30 days. Look at it in a simple UI, or in Grafana through Prometheus.

One binary · SQLite · alerts with browser push · MCP for AI assistants · no personal data

Quick start Source on GitHub
The agg Insights page: page views, purchases, revenue, visitors, average order value, bestsellers per category and top pages for the last 24 hours.

How it works

  1. Collect

    A 3 KB script sends page views and your own events, agg.track('feature_used', {…}), in batches, with retries, plus page, browser and device. Backends, cron jobs and CI send the same events with one HTTP request. You decide what to send.

  2. Aggregate

    Define what to compute, without code: count, sum, distinct count, last value or last time; grouped by any field; ranked inside groups. Values update as events arrive.

  3. Look

    A built-in UI with tiles, trends against the previous period, top lists and history. Or export to Prometheus and use Grafana and Alertmanager.

  4. Act

    Alerts when something happens or stops happening, as browser notifications, in-app or to a Slack / Discord webhook. A JSON API returns the values you mark public.

What you can compute

Websites

  • page views and visitors vs the previous period
  • top 10 pages today, this week, this month
  • top referrers
  • most read articles per section

Products and SaaS

  • feature usage and users per feature
  • active users in the last hour, day, week
  • sign-ups per plan and per source
  • sign-up rate, this week vs last week

Shops

  • purchases and revenue per product
  • bestseller in each category this week
  • visitors viewing a product right now
  • time since the last purchase, average order value
An aggregate
agg.track('purchase', { order_id, value,
  items: [{ id, name, category, quantity }] }, order_id)

Events     purchase
Explode    props.items
Operation  sum of item.quantity
Group by   category = item.category
Rank by    product  = item.id
Windows    5m 1h 6h 24h 7d 30d (+ previous)
Its values
GET /v1/top?site=pk_…&aggregate=bestsellers
           &window=7d&category=shoes

{ "items": [
  { "key": "73", "label": "Trail shoe", "value": 41 },
  { "key": "12", "label": "Road shoe",  "value": 27 }
] }

Simple UI, or Grafana + Prometheus

The built-in UI covers the everyday questions: what happened in the last hour, day, week or month, compared with the period before, and what is on top.

If you already run Prometheus, create a token-protected export and scrape it. Window values are gauges, all-time counts are counters, per-product series are limited to a top K or an allowlist. A Grafana dashboard and a docker-compose file are included.

agg_value{site="shop",aggregate="purchases",window="24h"} 123
agg_dimension_value{site="shop",aggregate="purchases",
  window="24h",product="73"} 5
agg_events_total{site="shop",aggregate="purchases"} 5012
agg_formula_value{site="shop",formula="aov"} 182.4

Set it up by talking to your AI assistant

agg has a built-in MCP server. Connect Claude or another assistant with a revocable token and ask it to plan your analytics: it picks a preset, tells you which events to send, checks what your site really sends, tests aggregates on real events, and creates them, with alerts, after you confirm.

Presets for websites, online shops, SaaS products and blogs work without an assistant too: one click when you add a site.

claude mcp add --transport http agg \
  https://agg.example.com/mcp \
  --header "Authorization: Bearer agg_api_…"

> Plan analytics for my SaaS: which features
  do paying users use? Alert me when
  sign-ups stop.

Privacy

Quick start

Run it one of three ways. The first start prints an admin token; the UI is on port 8080.

Docker

docker run -d --name agg -p 8080:8080 -v agg-data:/data ghcr.io/worotyns/agg:latest
docker logs agg        # shows the admin token

Fly.io

git clone https://github.com/worotyns/agg && cd agg
# edit app and AGG_PUBLIC_URL in deploy/fly/fly.toml, then:
fly apps create agg-example
fly volumes create agg_data --size 1 --region waw -c deploy/fly/fly.toml
fly secrets set AGG_ADMIN_TOKEN=$(openssl rand -hex 24) -c deploy/fly/fly.toml
fly deploy -c deploy/fly/fly.toml

One machine with a volume for the SQLite database; details in the README.

From source

git clone https://github.com/worotyns/agg && cd agg
go build -o agg ./cmd/agg   # Go 1.27+
./agg serve                 # prints the admin token, UI on http://localhost:8080

Log in, create a site and add its snippet to your pages:

<script async src="https://agg.example.com/agg.js" data-site="pk_…"></script>

Page views, visitors, top 10 pages and referrers work right away. For product insights, send events from your app and pick a template (feature usage, active users, sign-ups per plan…); the tester shows what will be counted on your real recent events before you save.

agg.track('feature_used', { feature: 'export_pdf' })

No browser needed: a backend, a cron job or a CI pipeline posts the same events over HTTP. An id makes a retried webhook count once.

curl -X POST https://agg.example.com/e -H 'Content-Type: application/json' \
  -d '{"site":"pk_…","events":[{"name":"invoice_paid","id":"inv-42","props":{"amount":49}}]}'

More examples, limits and headers in the HTTP API docs.

What it is not

It is not a full web analytics suite: no sessions, funnels, user profiles or session replay. It does not render widgets or pop-ups on your site; it gives you the numbers and an API to build them on.

It is source-available under the Elastic License 2.0: free to use, modify and self-host, also commercially; not to be resold as a hosted service.