Run these on us too. If we fail one, we would rather hear it from you than not hear it.
Astrology is hard to falsify. Astrology software is not. Underneath every one of these apps is arithmetic that is either right or wrong, and you can check it without believing a single thing about Saturn.
Here are six tests. They take about ten minutes each and they work on any app.
1. Does it ask for your birth time — and does it say what happens without one?
The ascendant moves through all twelve signs in 24 hours. Without a birth time, every house-based statement in the chart is a guess, and about half of the technique is house-based.
An app that takes your date and confidently produces a full house chart is either assuming a time (usually noon or sunrise) or making it up. Neither is disqualifying. Not telling you is.
What good looks like: the app says which parts are still solid without a time (Moon sign, nakshatra, most of the dasha sequence) and which parts it will not claim.
2. Check one ingress against a panchanga
This is the sharpest test there is, and it takes two minutes.
Find the app's date for a planet moving from one sign into the next — an ingress. Then look up the same ingress in an almanac you trust: Drik Panchang, or any published panchanga.
Positions are hard to compare by eye; ingresses are a single instant, so disagreement is unmissable. An app that is a third of a degree off will match on every screen and then miss an ingress by half a day.
We know, because that was exactly our bug, and a reader found it this way.
If they disagree by more than a few hours, one of them is wrong. Ask which.
3. Ask it what it does not know
Every astrology engine has holes. Every single one. The tradition is enormous, inconsistent in places, and nobody has finished it.
So: find the edge. Enter a chart with an unusual configuration and see whether the app ever says I don't know.
An app that has an answer for everything has an answer for nothing. It is generating text, not reading a chart. If the same paragraph structure appears for every sign with the nouns swapped, you are reading a template.
What good looks like: this one says "training needed" and shows you which Moon signs it has actually been researched on, and which it has not. Seven of twelve are still incomplete, and that is on the screen rather than hidden.
4. Cast the same chart twice, a week apart
Enter identical birth details on two different days and compare the reading word for word.
If it changed, something in there is random, or a language model wrote it fresh both times. Neither is astrology. A chart is a fixed set of positions; the reading of it should not drift because it is Tuesday.
Corollary: if the app has an AI chat feature, ask it the same question twice in two sessions. Watch what happens.
5. Look for the price of fear
Count how many screens tell you something is wrong with you, and then count how many of those screens have a button underneath.
The business model that pays best in this category is: name a defect, price the fix. Mangal dosha. Kaal sarp. Sade Sati. Each of them a real classical term, each of them stripped of its actual conditions and turned into a product.
What good looks like: an app that will tell you a hard thing and then tell you the true mitigating factor in the same chart, and sell you nothing. We have written down why we will never sell a remedy — hold us to it.
6. Ask where your birth data went
Open the app in aeroplane mode and cast a chart. If it works, the arithmetic is happening on your device.
Then read the privacy policy for the word "partners". Your birth date, time and place — the exact triple used for identity verification at half the banks on earth — is worth real money to a data broker, and a free astrology app has to be making money somehow.
What good looks like: computed on your device, and the app says so in a place you can check.
Run them on us
We mean this. Everything above is a test this app is meant to pass, and we have already failed one of them publicly and written it up.
If an ingress here disagrees with your panchanga, or a reading changes between two casts, or you find a screen that names a defect and offers to fix it for money — tell us. There is a real human address on every page of this Journal, and a disagreement you file inside the app goes into the research rather than into a support queue.
An instrument that cannot be checked is not an instrument. It is a mirror angled to flatter you, and there are already enough of those.