✠ ENTRY IV
The loop drained five build-plan phases (4 through 8) end to end while a much bigger parallel track ran alongside it: the owner shipped Themed Deck Library v3 straight to main (10 themes, 70 cards, 30 keywords, superseding the design-only spec 32 catalogue), landed a base-engine combat-difficulty patch, and opened a 21-agent card-rebalance PR that verify caught two real regressions in before it could merge. Critique's playtest-permission block is now a second post-fix recurrence (passes 5 and 6), still unresolved.
What changed
added population-level Vitest witnesses for the HP kill-paths still missing them (Conclusion/BODY-sig, Execute finish, escalation clock).
Why it changed
closes the last population-coverage gap the calibration phases (1-4) were chasing since nexus adoption.
commit d9763659
What changed
the mobile Village screen gained a SELL tab wired to the engine's existing sellItem / defaultSellPrice.
Why it changed
the engine has shipped a sell path since Phase 37 with no mobile consumer — a known plan/CURRENT-STATE.md gap.
commit 98e7b2bc
What changed
a read-back UI for past deaths (hazardDeathCount) and banked keepsakes, added to the Memoir screen.
Why it changed
presenter-only surface for state the engine already tracked but nothing rendered.
commit 48d2c802
combatMana sliceWhat changed
deleted the unused CombatManaState interface and rewrote four stale comments (actions.ts, combat-hud.engine.ts, StatusCard.tsx, DebugHudOverrides.tsx) still narrating a mana model that no longer exists.
Why it changed
verified the actual migration had already shipped nine days earlier (6ef5f989/160ae907) — the AUDIT row was carried into the monorepo re-onboard without being re-validated against code that had already moved on.
commit bdca1068
What changed
authored the three quests the map declared but never instantiated (get-to-forest, gather-wood, get-to-cave) and wired the missing collect objective type into resolveMapEvent.
Why it changed
the 10→25-node map expansion had already shipped pre-nexus; the real remaining gap was these three QuestName union members with zero Quest objects behind them.
commit 3bb6eeaa
What changed
removed in-combat retreat; the round-5 threat escalation now also intensifies enemy-inflicted status and grants a new enemy passive/curse every 5 rounds.
Why it changed
the base-engine half of the owner's response to the URGENT "combat challenge-gradient retune" phase candidate (measured win rate ≈ 1.00 across every stage) — raises the floor before the card-level rebalance in PR #41 lands on top of it.
What changed
replaced the v2 card catalogue with 10 self-contained themes × 7 cards each (70 total), a 30-keyword registry, 10 preset decks, and a schema-level strike-purge lint.
Why it changed
v2 had specced the strike purge but never enforced it — the root cause of recurring strike/chip sightings — and this supersedes the design-only spec 32 catalogue with a shipped implementation.
What changed
every Aporia labyrinth room now gets matched wall/door art (plus a follow-up cleanup pass), and the deck-preset picker switched from stance-colour swatches to keyword-based presentation.
Why it changed
visual/UX follow-through on the labyrinth ship and the new themed-deck system landing the same week.
What changed
added .claude/agents/card-expert.md — a card/keyword design consultant grounded in the Dawncaster KB (1,692 cards / 141 keywords) and the spec 32/v3 pricing doctrine, analysis-only like mechanics-expert.
Why it changed
gives card-design work (like PR #41's rebalance) a grounded consultant instead of ad hoc judgment calls.
commit dd2a50a9
What changed
enchant/disenchant cards were PAID-only persistent passives with no FREE line — playing one felt inert (their effects are subtle, level-10+ engine hooks). They now carry BOTH lines: the FREE (dieless) line grants a timed 3-round instance of the same passive; the PAID line makes the identical passive permanent (rest of combat, unique-in-play, leaves the deck cycle). A PAID play promotes a live FREE instance. Same effect and magnitude on both lines — only the duration differs.
Why it changed
owner-directed (2026-07-08) — the "weak buff now / commit the die for the whole fight" split makes these cards do something visible immediately, and gives the die a real decision instead of a mandatory tax.
✠ The marginalia
Each card's authored themed passive is unchanged — v4 only adds the FREE (timed, 3 rounds) line beside the existing PAID (permanent, rest of combat) one. The effect and its magnitude are identical on both lines; only the duration differs.
Enchantments — player-side positive passives (FREE → your tempZone; PAID → persistentZone):
| Card | Theme | Passive (same on FREE 3-round & PAID permanent) |
|---|---|---|
| Venom and Vein | Affliction | Every bleed/poison you apply lands at +1 intensity. |
| Practiced Cadence | Peroration | +1 Premise on the first card you play each turn. |
| Anvil of Form | Forge | Every kindled/floating die arrives with +1 pip. |
| Crown of Thorns | Akrasia | While Fallen, your status applications land at +1 intensity. |
| Achilles and the Tortoise | Control | Draw 1 each time you deny an enemy turn. |
| The Oracle's Eye | Oracle | The enemy's next stance is always revealed, plus omen riders. |
| Bone Orchard | Harvest | 1 HP drained from the enemy per Soul you gain. |
| Irresistible Grace | Charm | Your SWAY stops decaying and each new gesture lands harder (momentum). |
| Hedgehog's Dilemma | Bulwark | Every THORNS trigger also marks the enemy. |
| Resonant Chamber | Echo | The first spell each turn gains ECHO (fires twice). |
Disenchants — negative passives attached to the ENEMY (FREE → enemyTempAttachments; PAID → enemyAttachments):
| Card | Theme | Curse (same on FREE 3-round & PAID permanent) |
|---|---|---|
| Suppurating Curse | Affliction | Doubles the enemy's total DoT damage each round. |
| Captive Audience | Peroration | While you hold 4+ Premises, the enemy stays marked. |
| Entropy Tax | Forge | Every kindled/floating die you spend marks the enemy. |
| Mirror of Guilt | Akrasia | Every self-debuff you take lands 1 stack of it on the enemy too. |
| Quagmire of Doubt | Control | Enemy telegraphs enter play 1 rung lower. |
| Fated Course | Oracle | Every hit omen marks the foe. |
| The Tithe | Harvest | Enemy afflictions expire 1 turn sooner (faster Soul churn). |
| Mirror of Longing | Charm | Damage your defenses prevent becomes SWAY. |
| Crumbling Resolve | Bulwark | A fully-blocked attack costs the enemy a rung on its next telegraph. |
| Stuck in Their Head | Echo | Every ECHO / REPRISE drips 2 damage to the enemy. |
| Time (UTC) | Verb | Outcome |
|---|---|---|
| 2026-07-07 12:11 → 23:36 (~6 ticks) | march | all green; phases 4, 5, 6 shipped in sequence, plus critique passes 4 and 5 dispatched in-tick |
| 2026-07-07 17:34 | critique pass 4 | 0 findings — chronic playwright-permissions blocker persists |
| 2026-07-07 18:42 | fix (owner-directed, PR #32) | widened the unattended-run allowlist to include mcp__playwright__browser_*; marked pass 1-4's block "addressed" in plan/CRITIQUE.md |
| 2026-07-07 21:49 | critique pass 5 | 0 findings — identical block recurred even with the allowlist fix confirmed correct; re-opened as needs-user-call |
| 2026-07-07 19:08 | merge (owner, PR #40) | Themed Deck Library v3 — 10 themes, 70 cards, 30 keywords, straight to main |
| 2026-07-08 00:14 → 00:16 | march | all green, no-op tick |
| 2026-07-08 00:54 | feat (owner, direct push) | in-combat retreat removed; threat escalation intensified |
| 2026-07-08 03:48 → 03:56 | march | phase 7 CI-fallout fix (d6e89f96) + critique pass 6 dispatched in-tick |
| 2026-07-08 03:53 | critique pass 6 | 0 findings — same block, second post-fix recurrence |
| 2026-07-08 03:56 | feat (owner, direct push) | card-expert subagent added |
| 2026-07-08 06:29 | PR opened (owner, PR #41) | 21-agent thematic rebalance of all 10 preset decks; verify not yet run at open time |
| 2026-07-08 06:40 → 07:00 | march | phase 7 shipped |
| 2026-07-08 07:33 | CI (PR #41 branch) | verify red — two real regressions caught (see Needs you) |
| 2026-07-08 09:39 → 10:05 | march | phase 8 shipped |
| 2026-07-08 10:48 | night (this run) | digest — breadth check needed the same manual Playwright browser install as every prior night run |
Build-plan queue (plan/steps/01_build_plan.md): 9/15 shipped (phase 0 plus 14, 1-8) — four phases shipped since this morning (5, 6, 7, 8). plan/AUDIT.md 7 pending (down from 8 — the combatMana row drained itself as part of phase 7, moved to Done), plan/CRITIQUE.md 5 pending (same rows, now at pass 6 on the recurring one), plan/PHASE_CANDIDATES.md 8 pending (one row is a stale ~~...~~ SHIPPED 2026-07-05 marker still sitting under Pending instead of Promoted/Done; /expand still hasn't run since the 2026-07-03 seed). No open triage:needs-user or loop:do issues. Deploy gate: stub, exits 0 by contract (EAS wiring still queued for phase 11). Breadth check (npm run e2e:minigames, browser installed by hand): all four suites — hazard, combat, gathering, encounter-routing — ALL PASS.
None filed by digest this pass (in scope: loop-gate mistuning only — game-balance tuning routes through /expand or the domain tuning skills). The playwright-permissions recurrence above is a candidate for an /oversight mechanism decision, not a gate-constant tuning, so it stays in Needs you rather than becoming a plan/PHASE_CANDIDATES.md row.
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.