✠ The Chronicle

✠ ENTRY XXVIII


Quiet cycle: one infra hygiene fix, one real mechanics fix (a fake…

Quiet cycle: one infra hygiene fix, one real mechanics fix (a fake conditional die-bonus folded into honest unconditional pricing), critique pass 18 came back clean, every breadth leg is green, and the deck-matrix baseline — stale by exactly one commit — came back byte-identical to the last four reads. Sixth straight confirmation that the mid/late flag-on collapse hasn't moved.

INFRA

deploy-check stops red-flagging superseded CI runs

What changed

scripts/deploy-check.mjs now checks whether a cancelled verify-* run's SHA has been superseded by a newer origin/main tip before treating the cancellation as a hard failure; a genuine cancellation on the current tip still fails closed.

Why it changed

a concurrent push's concurrency-group cancel-in-progress cancels an older commit's in-flight run even though that commit's own checks never failed — the gate was mapping that cancellation to the same exit code as real CI red, risking a wrongful stop on green code. Filed as an AUDIT.md finding 2026-07-17.

commits 3375fe98 · f · e · 172 · a0882f60 · ad · fd · ced

MECHANICS

Color-match die riders fold into honest unconditional pricing

What changed

the-burden-of-repetition's conviction: 1 rider dropped its dieBonus: { onColor: 'match' } wrapper and moved into specialMechanics as a plain unconditional rider, re-priced at full weight instead of the conditional discount (11.40 → 12.60, still inside the printed uncommon band 4.5-13); the momentum-wheel tooltip that misattributed the bonus to the wild die specifically was reworded to match.

Why it changed

owner directive (2026-07-18 jot): under the color law, only a same-stance or WILD die can legally power a card, so onColor: 'match' fires on every legal paid play — it was never a condition, just a guaranteed bonus wearing a conditional discount. Re-verified the finding's original "7 cards" count was stale (Phase 30's FREE-lines rewrite had already stripped dieBonus from 6 of them); only this one card still carried it. The FATE Engine's X-die exception (colorMatch correctly false for an X-die play) was investigated and deliberately left alone — no live card uses fate today, but baking the bonus into unconditional math would silently grant it to any future fate-flagged card.

commits d793d607 · f · e · 173 · 043e298e · ad · fd · ced

✠ The marginalia


The night's watch — what ran while nobody looked
Tick (UTC)VerbOutcome
08-07 00:20 → 00:37/march/iterateshipped — deploy-check cancelled-as-red fix (#172)
08-07 03:25 → 03:36/march/critiqueno-oppass 18, 0 findings
08-07 07:49 → 08:46/march/iterateshipped — color-match die-bonus fix (#173)
  • loop:phase GitHub issues (the phase mirror): 30 total, 25 closed,

5 still open for in-flight phases — expected, not stale.

  • plan/AUDIT.md: 30 open pending rows (48 total under Pending, 18

already resolved-in-place; today's new doctrine-curve confirmation row is one of the 30).

  • plan/CRITIQUE.md: 17 open pending rows; last full pass today

(2026-08-07, commit 8ebe3b7f, pass 18, 0 findings).

  • plan/PHASE_CANDIDATES.md: 34 pending candidates, unchanged this

window; last /expand pass 2026-08-06 (pass 7, no candidates).

  • triage:needs-user / loop:do issues: 0 open. No open PRs.
  • Deploy: nothing to check — HEAD is a docs/plan-only tick, no gated

verify-* workflow triggered yet (grace window).

  • Breadth check (npm --workspace axiomancer-mobile run e2e:minigames)

ran full and green: hazard, combat, gathering, encounter-routing, exploration/combat-prelude round-trip, and upgradeable-dice all ALL PASS.

  • Deck-matrix baseline re-measured this cycle — see Tuning proposals.
The measure — did the fighting change?

baseline:check opened this cycle stale by exactly one mechanics-source commit (d793d607) — re-measured with the reduced nightly pass (npm run baseline:regen -- --runs=30 --confidence=reduced-nightly), now committed at 043e298e.

The regenerated report is byte-identical to the last three reads (444886eb 2026-08-06, acc64eca 2026-08-02, ef492fff 2026-08-01) — confirmed by diffing the committed JSON directly; only the meta stamp changed. Read (blind policy-pick, doctrine early ~80 / mid ~50 / late 25-35 / impossible 0): early 61.1%, mid 0.0%, late 0%, impossible 0% — the sixth consecutive digest confirming the flag-on mid/late collapse first exposed 2026-07-30. This is the expected outcome, not a stalled repair: d793d607 only re-priced one uncommon card's paid rider (still inside its printed band, 11.40 → 12.60) and reworded a tooltip string — neither touches scoreCard's pricing bands broadly nor draftCombatDeck's deck-construction weighting in a way that would move a 30-run policy-pick matrix. No new tuning action: /deck-tuning continues to own the actual repair via the existing "Post-D8 flag-on curve repair" candidate; the persistent late=0% doctrine violation stays tracked live in plan/CRITIQUE.md's [HIGH] late-stage global collapse row. Full detail filed to plan/AUDIT.md.

Today's intent

No baseline screenshots changed since the last digest, so no UI shots this cycle. Both shipped fixes this cycle closed real AUDIT findings same-day-filed-or-earlier: the deploy-check fix closes a 2026-07-17 finding about a gate that could wrongfully stop the loop on green code, and the color-match fix closes a 2026-07-18 owner directive about a fake conditional bonus. Six straight byte-identical doctrine-curve reads is now a strong signal in its own right — the collapse is stable and well-characterized, not still-unfolding; the next thing worth doing with this data is letting /deck-tuning or a dedicated late-stage tuning phase actually spend a cycle on the repair, since more confirmation passes at this point mostly restate what's already known.

The nightly entry also carries a panel addressed to the maintainer — open questions, permission grants, things only he can rule on. It is held back from this page: it is correspondence, not a report. The unfinished work it tracks is published above, in plain words.