| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| caprock_0.54.4_windows_amd64.zip | 2026-09-06 | 8.9 MB | |
| caprock_0.54.4_windows_arm64.zip | 2026-09-06 | 8.2 MB | |
| checksums.txt | 2026-09-06 | 600 Bytes | |
| caprock_0.54.4_darwin_amd64.tar.gz | 2026-09-06 | 8.8 MB | |
| caprock_0.54.4_darwin_arm64.tar.gz | 2026-09-06 | 8.4 MB | |
| caprock_0.54.4_linux_amd64.tar.gz | 2026-09-06 | 8.7 MB | |
| caprock_0.54.4_linux_arm64.tar.gz | 2026-09-06 | 8.1 MB | |
| README.md | 2026-09-06 | 770 Bytes | |
| v0.54.4 source code.tar.gz | 2026-09-06 | 6.3 MB | |
| v0.54.4 source code.zip | 2026-09-06 | 6.6 MB | |
| Totals: 10 Items | 63.9 MB | 0 | |
Fixed
- Half of Codex's token reports were being thrown away. About half the turns in a Codex transcript state how many tokens they used without breaking it down by kind, and Caprock read only the breakdown — so those turns were stored as having used nothing at all. On the owner's machine that hid 18.6M tokens and turned $23.71 of usage into $0.53.
Such a total is now counted as input, which is the unqualified rate: nothing is credited a cache discount it was not reported to have earned, and no split is invented across kinds. For those turns the cost is therefore an upper bound — any part of the total that was really a cached read cost a tenth of what is shown. A turn that genuinely used nothing is still stored as nothing.
