Repository navigation
sitemap.xml stamps the build date as lastmod on all 53 pages, up to 4.5 months of freshness the pages do not have #181
Description
Measured by fetching https://sprout.chelseakr.com/sitemap.xml and the pages it lists on 2026-09-13.
All 53 of 53 sitemap entries carry the same <lastmod>: 2026-09-13. That is today — the date the docs were built. Tomorrow's build will say tomorrow, on all 53, whether or not a single page changed.
The pages themselves carry their own dates, and they do not agree:
| URL | sitemap lastmod |
newest date visible on the page |
|---|---|---|
/ADAPT/ |
2026-09-13 | 2026-05-01 |
/ARCHITECTURE/ |
2026-09-13 | 2026-06-22 |
/PUBLICATION-READINESS/ |
2026-09-13 | 2026-08-16 |
/RESEARCH-ROADMAP/ |
2026-09-13 | 2026-07-16 |
/RESPONSIBLE-TECH-AUDITS/ |
2026-09-13 | 2026-07-12 |
A document last touched on 2026-05-01 is advertising itself as modified today — four and a half months of freshness it does not have.
lastmod is the one field in a sitemap that makes a factual claim about the content. Here it is a claim about the CI runner. It is the same class of defect as an "absence rendered as a value": a value is being published where the real answer is "we did not measure this per page".
Two ways to fix it
- Derive it per page. mkdocs can emit a real per-page modification date (
mkdocs-git-revision-date-localized-plugin, or a post-build step that readsgit log -1 --format=%cI -- <source>for each page's source file). gtfs-scorecard already does the honest version of this — a spot check of six of its agency pages foundlastmod 2026-09-13matching the "checked 2026-09-13" printed on each page. - Drop
lastmodentirely. A sitemap with nolastmodis valid and says nothing false. Four other sitemaps in this portfolio omit it. Absent beats wrong.
Either is an improvement. Keeping the build date is the only option that publishes something untrue.
No source change is proposed here and no PR is open; this is a report of what the live host returns.
Prepared with AI assistance; reviewed before submission.
Activity
ChelseaKR commentedon Sep 13, 2026
Premise re-verified: real, still live, and already claimed by #182
Checked independently today from a clean worktree off
origin/main(dc85850) with a fresh venv, and against the deployed file.mkdocs build --strictonmainwritessite/sitemap.xmlwith 53<loc>, 53<lastmod>, all2026-09-13.https://sprout.chelseakr.com/sitemap.xmlreturns the same: 53 of 53 entries reading2026-09-13.#182 is open and closes this, and it reproduces exactly: 52 URLs with a defensible
lastmod/ 53 URLs examinable, across 16 distinct days from2026-06-22to2026-09-07, none of them the build's own date; the single omission is the site root, deliberately. Its headline negative control was re-run from scratch here and fires — removingtheme_overrides/sitemap.xmlaborts the build under--strictand names the build date it would otherwise have published on all 53. The detail is in a comment on that PR.No second branch was opened. This stays open until #182 merges, which will close it.
Prepared with AI assistance; reviewed before submission.