Every claim on this page is something a module refuses to load without. These are not tests we run; they are conditions of importing the code.
A published patrika, value by valueThe computed tables are checked against a published Gujarati janma patrika — the particulars, the vargas, the bhava table, the upagrahas. Where our figure and the published page disagree, the disagreement is pinned by a test rather than tolerated, so it cannot quietly grow.
Ashtakavarga reproduces the published chart exactlySign by sign, not just row totals. A wrong house passes a totals check — that exact typo shipped here once, and only a per-sign comparison against a real chart caught it.
Shad bala, all twenty sub-rules asserted separatelyShad bala is a sum of twenty sub-rules, and two errors of opposite sign cancel while the page still looks right. So every one of the twenty rows is asserted against the reference's own printed intermediate figures, not just the total.
KP on the Krishnamurti ayanamsha, not LahiriAbout 5'48" apart in 2025 — small enough to look ignorable, and it is not: on our reference chart the two land Jupiter in different subs of Punarvasu. Sub-lords are computed on Krishnamurti's own ayanamsha and Placidus cusps.
Jaimini karakas, both schoolsThe chara karakas with Rahu counted and without — seven and eight — because the reference prints both columns and settles nothing about which school its software follows. The rashi drishti table reproduces all twelve rows.
Lal Kitab against three published reportsTwo independent charts with different lagnas, eighteen placements, the fixed house tables three times over and every antardasha boundary of the first 35-year cycle.