Public, live status report on the website

The static Compatibility status page (landed 2026-07-15) is hand-written prose: it describes what compiles (c-testsuite, zlib, SQLite, Lua, tcc, cJSON; the RTL suites; fpjson/Synapse/fgl; the conformance snapshot) with the correct claims discipline (the two distinct "byte-identical" meanings, never "clone"). Good as a narrative, but the numbers rot — they are a manual snapshot of gates that move every day.

What already exists

tools/twatch_web.py --static --out DIR writes self-contained dashboard.html, bench.html, and conformance.html into tstate/, fed by the watcher's per-SHA conformance.tsv / <host>.json / bench rows. So the data pipeline and renderers already exist — Track T publishes them to tstate/.

The gap

Those generated pages live in tstate/ (the watcher's write scope) and are not surfaced on the public website, which publishes docs/** verbatim. So:

1. Publish path — expose the generated conformance/bench/dashboard pages (or a curated subset) at a stable public URL, linked from docs/reference/status.md.
2. Freshness — either (a) link out to the live generated report and keep status.md as the stable narrative, or (b) generate the corpus-matrix section of status.md from the testmgr/tstate data so the counts self-update.

Cross-track

Track D owns the website/publishing and the prose; Track T owns the generators and the tstate/ data. A clean split: T emits a machine-readable status artifact (counts per corpus/suite + last-green SHA); D renders/links it. Keep the claims discipline in whatever is auto-generated — the "output parity vs self-host reproducibility" distinction must survive templating (see the compatibility-claims note in the agent guide).

Not doing yet

Exploratory (user: "we may seek a way"). Filed so the static page and the existing tstate renderers can be joined up when the website build is ready for it.

RESOLVED 2026-08-30 — premise discharged, not built

This idea's own argument no longer holds, verified by frankD before it built anything on top of it. The idea was filed when status.md was hand-written prose, and its case is "the numbers rot — they are a manual snapshot of gates that move every day."

That stopped being true in ce89ff14b, "docs(D): stop hand-writing numbers in status.md." The page now opens "For current numbers, read the live status pages, not this page" and states outright: "Deliberately no figures — a number written here is a number that starts going stale the day it is written." That is this idea's own option (a), already shipped. Gap items 1 and 2 are both closed.

Its second concern — that claims discipline must survive templating in anything auto-generated — is also nothing to do, and this was grepped rather than reasoned: tools/twatch_web.py and tools/testmgr.py carry no byte-identical / clone / compatible / parity language at all. The generated pages carry numbers; the claims live in docs/** prose; that prose was audited against CLAUDE.md's claims-discipline table on 2026-08-29. frankD expected this to be a real finding and reported that it was not — "I'd have reported it as one if I'd reasoned instead of grepped."

What the fix EXPOSED, which is the only live work here

Carried forward as task-d-verify-the-published-status-urls-docs-now-delegates-all-numbers-to [D p30]:

A better trade, not a free one, and never checked once. docs/** references six distinct pxxc.org/status/* URLs across two files; tools/ has no link checker of any kind (docaudit checks internal citations, docsnip compiles code blocks, neither looks at an http URL); and the generated HTML is not in this checkout, since twatch_web --static writes it inside the watcher's clone — so there is no local artefact to diff either.

The sharp version, frankD's: status.md does not merely link to those URLs, it delegates its entire numeric content to them. If /status/conformance/ is missing, the public compatibility page has no figures and no working pointer to any, while telling the reader that is exactly where the figures are — strictly worse than the snapshot it replaced, and the one outcome nobody can notice from inside the repo.

Resolved by the coordinator rather than by the lane, because closing an idea is a filing decision and frankD correctly declined to claim a ticket it had been told not to claim.