Since May 1
May 1 – Oct 26 · out-tokens per day9,632Sessions
1.1MLines added
328.9MTokens out
Peak: Oct 11 — 770 sessions in one daycoolhot
Sheet B-01 — telemetry
Meter readings, in full.
Everything on this site is built with Claude Code. Parsed locally from 3,763 session logs, May 22 – Oct 26, 2026.
1.1Mlines addedacross every repo, 158 days
9,648sessions23,868 prompts · 466,618 replies
328.9Mtokens writtenmodel output
85.4Btokens read from cachecontext served from the prompt cache
Yard ledger
01cubby8 storeys2,613 sessions · 251,797 lines · 95.8M tok out
02swatkats6 storeys732 sessions · 146,268 lines · 28.9M tok out
03claude-skills5 storeys1,214 sessions · 109,244 lines · 22.3M tok out
04folix5 storeys412 sessions · 101,792 lines · 31.0M tok out
05doc-scan3 storeys323 sessions · 41,587 lines · 15.9M tok out
06purix2 storeys125 sessions · 34,483 lines · 7.0M tok out
07floorprint2 storeys266 sessions · 32,064 lines · 8.2M tok out
08cartoon2 storeys188 sessions · 31,581 lines · 5.0M tok out
09foundry2 storeys397 sessions · 27,062 lines · 12.9M tok out
10raja2 storeys142 sessions · 23,265 lines · 6.4M tok out
11sift2 storeys140 sessions · 15,695 lines · 3.7M tok out
12memekit1 storey49 sessions · 9,855 lines · 842,696 tok out
13design-system1 storey16 sessions · 3,129 lines · 586,810 tok out
cubby — tallest hall on the yard, 251,797 lines and counting.
Per-repo figures are floors; old transcripts rotate out.
Top tools
Bash153,964
Read45,001
Edit18,560
Write8,398
StructuredOutput5,041
Model mix
466,618 assistant messagesSonnet 5 · 26%Opus 5.5 · 18%Sonnet 5.5 · 17%Opus 4.8 · 12%Fable 5 · 8%Opus 5 · 8%
146,302 thinking blocks3,263 subagent runs679 skill invocations/compact × 97longest prompt: 290,091 chars1116 screenshots into cubby
Counted from 3,763 local session logs by a claude-skills script — on-device, like everything else here. Regenerated at each deploy.