curl -s api.midgarcorp.cc/v1/ops/public/stats

Midgar, in the open

How many people use Midgar's apps, published as it is.

Counts, not people. An install reports that it was active, never whose.
No identifier leaves a device. No user, device or Midgar ID rides along, and no IP address is kept.
Anything under 10 shows as <10, so no figure points at a handful of people.
The code is open. Every event is declared in one public registry.

Updated 2026-10-30 10:11 UTC · weeks start on Monday, UTC

open stats --weeks 12

All apps

The last 12 weeks, newest on the right. A dashed end means the week is still running. Active and new installs count only app versions that send usage counts, which most apps begin with their next release; purchases and AI requests cover every app.

Weekly active installs

12 week of Oct 19

Installs that opened an app during the week.

New installs

<10 (hidden: fewer than 10) week of Oct 19

Every week is under 10, so none is shown. Installs first seen that week.

Purchases

<10 (hidden: fewer than 10) week of Oct 19

Every week is under 10, so none is shown. Purchases recorded that week.

AI requests

<10 (hidden: fewer than 10) week of Oct 19

AI chat requests answered that week.

every week and month as a table

Portfolio numbers for each of the last 12 weeks
Week of	Active installs	New installs	Purchases	Refunds	AI requests	Subscriptions
Oct 19	12	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Oct 12	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	14	<10 (hidden: fewer than 10)
Oct 5	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Sep 28	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Sep 21	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Sep 14	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Sep 7	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Aug 31	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Aug 24	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Aug 17	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Aug 10	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Aug 3	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)

Portfolio numbers for each of the last 6 months
Month	Active installs	Purchases	Refunds	AI requests	Subscriptions
Sep 2026	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Aug 2026	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Jul 2026	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Jun 2026	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
May 2026	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)
Apr 2026	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)	<10 (hidden: fewer than 10)

open stats --by app

Each app

Weekly active installs per app, the busiest first.

Weekly active installs of each app over the last 12 weeks
App	Last 12 weeks	Latest week
Flaccy		12
Dream Eater		<10 (hidden: fewer than 10)
Embr		<10 (hidden: fewer than 10)
Helia		<10 (hidden: fewer than 10)
Psywave		<10 (hidden: fewer than 10)
Solar Beam		<10 (hidden: fewer than 10)
Tailscode		<10 (hidden: fewer than 10)

cat registry/signals.yaml

What is counted

The apps can send 15 kinds of count and nothing else. Each carries only the app, the platform, the app version and, for some, the week the app was installed. The server keeps daily totals, never a log of what an install did, and the country read at Cloudflare's edge stays in identifier-free data points for 90 days.

Master of Flags sends no usage counts: it is in Apple's Kids Category. Crucible sends none either. Read Master of Flags' privacy policy.

the 15 counts the apps can send

app.launch not sent yet: The app was opened.
app.active: Daily, weekly and monthly active installs.
onboarding.complete not sent yet: The first-run introduction was finished.
paywall.view: How often the purchase screen is shown, and from where in the app.
paywall.purchase: Purchases completed in the app, by product.
paywall.checkout: How a purchase you started ended, and for which product.
funnel.step: First-run steps before a purchase.
portal_open: The screen with other Midgar apps was opened, and from where in the app.
impression: Another Midgar app was shown on that screen.
tap: Another Midgar app was tapped on that screen.
loaded: The App Store page of another Midgar app finished loading from that screen.
launch: Another Midgar app was opened from that screen.
overlay: The App Store sheet for another Midgar app was shown at a fitting moment.
metrics.daily: Daily numbers on how fast the app starts and how much memory it uses, so slow versions get fixed.
metrics.diagnostics: Counts of crashes, freezes and heavy disk or processor use, so they get fixed.

how each number is defined

Daily active installs: How many installs open an app on a day. Counts installs, not people; one person on two devices counts twice. A week or month reads as the average day.
Weekly active installs: How many installs open an app in an ISO week. Counts installs, not people. Each install reports once per week, so a week still running is a partial count.
Monthly active installs: How many installs open an app in a calendar month. Counts installs, not people. A month still running is a partial count.
New installs: Installs active in the week they were installed, the closest Lifestream gets to downloads that open the app. Only installs that ran a version with Lifestream and left usage counts on. App Store downloads are the downloads metric.
Week-1 retention: The share of a week's new installs that are active again the next week. The period is the install week. A cohort under 5 installs is hidden, and the latest cohort has no week 1 yet.
Week-4 retention: The share of a week's new installs still active four weeks later. The period is the install week. A cohort under 5 installs is hidden, and cohorts younger than four weeks have no value.
Paywall views: How often a paywall is seen, by placement. Each install counts once per day per placement.
Checkouts: What happened when a paywall's buy button was pressed, by outcome. The outcome is the app's view of the purchase; the ledger's purchases are the record of what Apple charged.
Paywall purchases: Checkouts that ended in a purchase. The app's view of the purchase.
Paywall conversion: Purchases per paywall view. Views are once per install per day per placement, so this is purchases per paywall day rather than per showing.
First-run steps: How many installs reach each first-run step, once per install. Each install reports a step once in its life, on the day it first reaches it.
Portal impressions: How often the cross-promotion portal or a contextual overlay shows another Midgar app, by target app and placement. Counts showings, not installs.
Portal taps: Taps on another Midgar app in the portal, by target app and placement. A tap opens the App Store page; the install, if any, is the downloads metric's campaign.
Sales: App Store sale events as Apple notifies them, by kind. Production only. Kinds are INITIAL_PURCHASE, RENEWAL, NON_RENEWING_PURCHASE, REFUND and REVOCATION as the ledger maps them.
Refunds: Purchases Apple refunded or revoked. Production only; the day Apple told the ledger, not the day of the purchase.
Active subscriptions: Subscriptions paid up at the end of each day. Production only, counted from the transactions the ledger has; a subscription bought before the ledger existed counts from its first renewal the ledger saw. Every app that ever sold a subscription has a value for every day, zero included.
Introductory offers started: Subscriptions that started on an introductory offer, a free trial in most apps. Production only. Apple's offer type 1 covers free trials and pay-up-front or pay-as-you-go introductory prices alike.
AI requests: Chat requests mako served, by model alias. Chat requests only; image generations are not counted here.

git clone github.com/guitaripod/midgarkit

Check it yourself

The counting is open source.

MidgarKit Signals ↗ what an app sends, on the device
signals-catalog.json ↗ every event and allowed value
metrics-catalog.json ↗ every metric and its formula
public stats JSON ↗ the numbers on this page

Something looks wrong or unclear? Write to support@midgarcorp.cc.