One of our six verdict lamps has never been red, and it cannot be. Every audit page opens with the same row of six, and across our 26 default audits those 156 lamps were set by 145 individual flags. 23 of the 26 carry at least one red lamp. Not one of the 26 reads ok on all six. The regime lamp reads red on none of them, and that is not a finding about the robots but a property of the three rules that can set it. This article is about what a lamp is, so that a green one is read for what it says and not for what it looks like.
Our engine knows four severities and ranks them in that order, ok below info, info below caution, caution below red. Each of the six dimensions starts the audit at ok. Every rule that fires attaches a flag to one dimension, and the dimension keeps the highest severity it has been handed.
A lamp is therefore a maximum and never an average. One red flag among six findings makes the dimension red, and five quiet findings beside it change nothing. The reverse holds too. A dimension that collected four info flags stays at info, however many of them there are. Counting flags and reading lamps are two different readings of the same audit, which is why the flag list sits open on every audit page under the lamp it produced.
Because the level only ever rises from its starting value, an ok lamp means exactly one thing. No rule in that dimension fired. It does not mean the dimension was examined and found healthy, and it carries no statement about anything our rules do not look for.
CryonX is the clearest case in the catalogue. It carries 2 flags in total, the fewest we have together with Gold Snap, and four of its six lamps are ok as a result. The one flag that matters is red and sits on prop fit. Four quiet lamps and a loud one is a common shape here, and the quiet ones are quiet because nothing was raised, not because something was cleared.
The other end of that range is Logan with 10 flags, the most in the catalogue. Its second place is held by an audit we keep unlisted, Pulse Engine, with the same count. Ten findings do not make ten lamps. Logan still ends on three red lamps and two ok ones.
Across the 26 audits the lamps distribute as follows, ok first and red last.
That is 46 ok lamps, 27 info, 42 caution and 41 red. Costs is the quietest dimension at 14 ok, which is worth holding next to the fact that our commission retrofit takes real money out of 22 of these 26 runs. A quiet costs lamp means the cost share stayed under our threshold, not that costs were small.
The regime dimension has three rules and their severities are fixed in the code. A long beta suspicion is a caution. Three or more negative years is a caution and two is an info. A single year carrying more than 60 percent of total profit is an info. None of the three can produce a red, so the maximum a regime lamp can ever reach is caution.
The measured distribution matches that ceiling exactly. Regime reads caution on 14 of the 26, which is more than half, info on 7 and ok on 5. Anyone reading the six lamps as a scale of danger would rank regime as the safest dimension in the catalogue, and they would be reading the rule set rather than the robots.
We publish this rather than quietly rescaling because the fix is not obvious. Making one of the three rules red would move a lamp on a large share of the catalogue at once, which is a decision about our own thresholds and not a measurement. Until that decision is taken and dated, a green regime lamp is the weakest green we print.
Prop fit reads red on 15 of the 26 audits, more than any other dimension, and it never reads info because its rules only produce caution and red. A red here means a run breaks a written rule of at least one of the three prop firm profiles we ship. On 4 September the red rules were three. An account that still hits the drawdown floor at one eighth of the tested size, a consistency limit crossed in more than two years, and positions held overnight or across a weekend where the profile forbids it. Since 29 September a fourth applies: no tested size stays inside the daily loss limit.
Prop Firm Gold EA is the name that makes the point. It is profitable in our run and its prop fit lamp is red anyway, because profit and rule compliance are separate questions. A trader with no prop account can read that lamp as noise. A trader with one should read it before the profit line.
Gold House and Gold Snap carry no red lamp at all, and so does one unlisted audit, Market Anomalies EA. That is 3 of 26. The remaining 23 carry between one and three, and the three that carry the maximum of three are Lizard, OilVector X and Logan.
Absence of red is not silence. Our engine holds a summary sentence for a run that raised no red and no caution flag anywhere, and it reads that no structural red flags were found. When we wrote this on 4 September, it had never been printed: not one of the 26 default audits qualified for it. Umbra Vault AI EA, audited later, is the first that does. Gold House and Gold Snap both reach their best result with cautions still standing.
Every count above is public in the audit files linked in place, and the category pages show the lamps without a purchase. Holding your own tester report? The browser check is free.