The Budget
Two different budgets govern this system, and conflating them makes you optimise the wrong one. That mistake has been made here, at length.
Standing context — present before the GM reads anything
| Always present | Tokens |
|---|---|
| The Ladder (pinned) | ~4,200 |
| 20 narrative triggers, evaluated on their phase | ~1,900 |
| Card - The Branch — declares "every turn anyone acts" | ~1,900 |
| Card - Voice — declares "at RUNG 4, every turn" | ~1,800 |
| 3 partial files, descriptions only | ~155 |
| Standing subtotal | ~10,000 |
Craft documents no numeric cap. It says only keep the pinned set small. An earlier version of this chapter asserted a hard 4,000-token ceiling as if it were platform law; it was invented, then obeyed for months while real costs went unmeasured. The working ceiling is ~6,000. The true constraint is attention: every standing token takes weight from every other one.
Triggers are not free
Twenty narrative triggers are evaluated every turn on their phase. The platform has no keyword matching — all of them are present whether or not they apply. That is why every trigger body opens with its own gate and says if not, ignore this.
Two cards are standing files in everything but name
They declare themselves per-turn in their own descriptions. Watch them exactly as you watch the Ladder, and apply one test before adding anything to either:
Does this apply on a turn where two people are simply talking?
If not, it belongs in a card that loads when it does.
Read cost — separate, and much larger
What the GM reads lands in the conversation window, not the standing set. Reads are cheaper than they feel but not free: everything read this turn shares room with the conversation itself. One card is comfortable. Three is a flooded window and a GM that has forgotten the last exchange.
Never read a whole mechanic. They run 3,000–26,000 tokens. That is what cards are for.