Two corrections to the pending entry, both mine.
The corpus line said the texts were written 'before generative models were in
use'. GPT-3's API was in beta from mid-2020, so that is not true of a 2021
revision; the claim that holds is 'before 2022', which is weaker and correct.
It is the wording the project itself uses in Docs/CALIBRATION.md.
And the entry counted tools -- 'ten in all', 'eight of the ten'. An entry that
enumerates goes stale the next time one ships; one that characterises does not.
This now says 'every tool but two runs entirely on the machine', which stays
true without anybody editing it.
Also adds the install command, since the package is on NuGet.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015PEbbiYSNPw7jE3LrPNhyF