✠ The Chronicle

✠ ENTRY XXIV

A quiet infra night — two long-standing gaps closed at the root: the…

A quiet infra night — two long-standing gaps closed at the root: the night runner's cold-Playwright breadth-check failure (ten nights running) and docs/combat.md's description of a screen that was deleted two months ago. Both landed clean, all six breadth-check legs pass, and the deck-matrix baseline is still fresh — no re-measure needed. The Closes #N no-op keeps recurring (thirteenth confirmed instance) and stays unpicked-up. Pulse review also caught and closed a stale duplicate issue.

INFRA

Night workflow now installs Playwright before the breadth check

What changed

.github/workflows/night.yml now passes install_playwright: true to the shared skill workflow, mirroring critique.yml and its peers.

Why it changed

without it, /digest's e2e:minigames breadth check aborted on every cold Actions runner ("Executable doesn't exist at .../chromium_headless_shell-1228/..."), confirmed unfixed for ten consecutive nights — each one needing a by-hand npx playwright install chromium-headless-shell workaround before the check could even run. plan/AUDIT.md finding [3.2], issue #165.

commits ca5b86f1 · e · 165 · — · ee · e · Ce · #N` · cad · be · f · why · this · eeded · a · adce · aa

INFRA

combat.md rewritten for the current Hazard-Pattern Combat screen

What changed

replaced docs/combat.md's description of the fully-retired legacy combat screen (deleted app/(tabs)/combat.tsx, the removed resolveCombatRound four-phase loop, a deleted SkillConfirmOverlay, three e2e paths that no longer exist) with the real <CombatEncounterPanel> / <CombatBoard> architecture — entry points, data flow, the CombatEncounterPhase union, the drag-to-power turn flow, board layout, CombatViewModel shape. Spec 04 marked superseded so it isn't mistaken for current architecture.

Why it changed

plan/CRITIQUE.md's standing "Engine doc-drift is chronic" / "Wrong-engine mental model in docs" rows (external-critique, score ~4.9) — a fresh reader following the old doc would have built a mental model of a screen that hasn't existed since 2026-06.

commits 6690c5d0 · e · 164 · ef · `npm · run · ef · ace · aacebe · 268 · e · / · 2715 · e · ee

INFRA

Thirteenth Closes #N no-op — still unpicked-up

What changed

#165's shipping commit (ca5b86f1) carried a correctly numbered Closes #165 trailer on a direct push to main — a workflow-config-only diff with no verify gate to apply — and still didn't auto-close the issue; closed by hand same-session.

Why it changed

the same unexplained symptom tracked since #83 (Phase 32 Part 1, 2026-07-13). Thirteen confirmed instances now (#83, #151, #155, #129, #156, #157, #158, #159, #160, #162, #163, #164, #165), no new root-cause signal this time either. Already escalated to the top of plan/PHASE_CANDIDATES.md (score 8.5, [score 8.5, escalated]) via /expand — this entry is a log update, not a new proposal; the loop doesn't get to fix its own constraint-tracking without /oversight.

✠ The marginalia

The night's watch — what ran while nobody looked
Tick (UTC)VerbOutcome
08-02 14:04 → 14:22/march → /iterateshipped — combat.md rewrite (#164)
08-02 19:40 → 19:45/marchno-op
08-03 04:03 → 04:07/marchno-op
08-03 10:10 → 10:27/march → /iterateshipped — night.yml Playwright fix (#165)
  • Build plan: 57/57 checked, none blocked.
  • plan/AUDIT.md: 31 open pending rows (42 total under Pending, 11

already marked resolved-in-place).

  • plan/CRITIQUE.md: 21 open pending rows; last full pass 2026-07-31

at commit 9a445281 (three days stale — next /critique due).

  • plan/PHASE_CANDIDATES.md: 34 pending candidates.
  • triage:needs-user / loop:do issues: 0 open (closed the one

stale duplicate below).

  • Deploy: nothing to check — HEAD is a docs/plan-only tick, no gated

workflow triggered.

The measure — did the fighting change?

None. npm run baseline:check reports FRESH — baseline acc64eca, measured 2026-08-02, confidence reduced-nightly, no mechanics-source commits since. No re-measure triggered, so no doctrine-curve bands to report against this cycle.

Today's intent

Breadth check (npm --workspace axiomancer-mobile run e2e:minigames, the project's real substitute for this skill's stale npm run e2e pointer — plan/AUDIT.md's standing divergence finding) ran full and green: hazard, combat, gathering, encounter-routing, exploration/ combat-prelude round-trip, and upgradeable-dice all ALL PASS. No baseline screenshots changed since the last digest, so no UI shots this cycle.

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.