What
Cut the release that lets an installing visitor experience what the docs teach, then carry it to the audience. The deployed docs site teaches post-0.4.1 behavior, but the packages a visitor installs are June 26's 0.4.1 — 40+ commits behind — and docs-taught input silently loses content on them (<code-block> renders empty, fixed on main by #379; numbered remark/proof cross-refs error, fixed by #374). A release is currently blocked: main's manifests depend on @enscribejs/ehtml ^0.4.1, which is unpublished (404 — the July 3 rename outran the registry), and all three workspace versions still read the already-taken 0.4.1. Gated on decision D4 (0.4.2 vs 0.5.0 — the audit leans 0.5.0 for a clean "this is what reviewers saw" baseline; Ariel's call).
Source: presentable-for-feedback stocktake audit, 2026-07-08 (the full report is a process artifact outside the repo). Execution: one CC slice closes this issue once D4 is decided; the checkboxes tick in that slice.
What
Cut the release that lets an installing visitor experience what the docs teach, then carry it to the audience. The deployed docs site teaches post-0.4.1 behavior, but the packages a visitor installs are June 26's 0.4.1 — 40+ commits behind — and docs-taught input silently loses content on them (
<code-block>renders empty, fixed on main by #379; numbered remark/proof cross-refs error, fixed by #374). A release is currently blocked: main's manifests depend on@enscribejs/ehtml ^0.4.1, which is unpublished (404 — the July 3 rename outran the registry), and all three workspace versions still read the already-taken 0.4.1. Gated on decision D4 (0.4.2 vs 0.5.0 — the audit leans 0.5.0 for a clean "this is what reviewers saw" baseline; Ariel's call).enscribe.orginto every deploy (static.yml:56) and that domain does not resolve; harmless today (Actions-based Pages ignores the CNAME file) but a live 301-to-dead-domain landmine if the custom domain is ever set in Settings → Pages. Fix: confirm Settings → Pages has no custom domain configured; drop the CNAME step until DNS exists; correct the workflow comment.@enscribejs/ehtml(first publish under the new name), bump all three workspace packages past 0.4.1 per D4, publish@enscribejs/enscribe+@enscribejs/cli.npm deprecatethe four legacy 0.1.0 packages (@enscribejs/core,/remark,/interpreter,/jats-import) plus@enscribejs/layer1-vocabulary, each pointing at the new names.npm installthe newly published packages and run the README taste example through the published CLI: the citation must render (per the front-door-truth fix) and docs-taught<code-block>input must not lose content. This is the test that fails on 0.4.1 today.Source: presentable-for-feedback stocktake audit, 2026-07-08 (the full report is a process artifact outside the repo). Execution: one CC slice closes this issue once D4 is decided; the checkboxes tick in that slice.