Andrew Fribush

fig. 2.9.3: resident context before and after on the two corner projects and the off-corner one, in tokens

fig. 2.9.3 · resident context before and after on the two corner projects and the off-corner one, in tokens.
THE PROJECT OUTSIDE THE VALIDATION CORNER: ABOUT 66%ROW = PROJECT; RING = TOKENS BEFORE, DOT = AFTER; ACCENT = OFF-CORNERRESIDENT CONTEXT, TOKENS05K10K15KPROJECT PIN THE CORNER · FOUR CONVENTIONS HELD AT ONCE · 66 ENTITIES3,25416,633~80% CUTPROJECT IIN THE CORNER · FOUR CONVENTIONS HELD AT ONCE · 33 ELABORATIONS1,59814,588~90% CUTPROJECT GOFF-CORNER · NO SESSION SUBJECTS · 128 OF 144 FILES EXTENSIONLESS · THE.md WALKER WOULD HAVE SEEN 15 OF THEM3,4459,925~66% CUT

Resident context is the always-loaded surface in tokens before and after, measured on each project’s dogfood branch: a full dogfood with 66 entities on Project P, summary-only projections on Projects I and G. Token counts and cuts are the source’s approximate (~) figures. Project G’s migration carried 139 files, additive (+1,112 lines, 0 removed).

source: cairn DESIGN.md, lines 68-69, 672, 713-714 and 867-876 (repo 12dd72d) · offCornerCut.data.ts

View in context: Counting the assumptions a test shares with its code → · All figures

Andrew Fribush, software and computer engineer in Austin. I build consequential systems where software meets institutions: finance, healthcare, commerce, government data, and the mechanisms that govern AI agents. Taking new engagements: hello@andrewfribush.com.