W85 — The Resort Tells You What Needs Attention

.claude/verify-report.json

@@ -1,18 +1,18 @@
 {
   "gates": [
-    { "name": "Typecheck", "status": "Passed", "detail": "npm --prefix src/engine run typecheck — exit 0" },
-    { "name": "Lint", "status": "Passed", "detail": "npm --prefix src/engine run lint — exit 0" },
-    { "name": "Test", "status": "Passed", "detail": "npm --prefix src/engine test (vitest) — Test Files 79 passed (79), Tests 1127 passed (1127)" },
+    { "name": "Typecheck", "status": "Passed", "detail": "npm --prefix src/engine run typecheck — tsc --noEmit && tsc --noEmit -p tsconfig.scripts.json — exit 0" },
+    { "name": "Lint", "status": "Passed", "detail": "npm --prefix src/engine run lint — eslint src scripts — exit 0" },
+    { "name": "Test", "status": "Passed", "detail": "npm --prefix src/engine test (vitest run) — Test Files 79 passed (79), Tests 1146 passed (1146)" },
     { "name": "Pack package", "status": "Passed", "detail": "cd src/engine; npm pack --silent — produced the-running-dev-game-engine-0.10.0.tgz, exit 0" },
-    { "name": "Inspect tarball", "status": "Passed", "detail": "tar -tzf the-running-dev-game-engine-0.10.0.tgz: no src/, no tsconfig*.json, no .test.* entries; package/dist/ present" },
-    { "name": "Consumer smoke", "status": "Passed", "detail": "consumer-smoke: cleared node_modules/package-lock.json/dist, then npm run install:engine && npm run build && npm run smoke — all three exited 0" },
-    { "name": "Parse-check PowerShell scripts", "status": "Passed", "detail": "Parsed every tools/*.ps1 with [System.Management.Automation.Language.Parser]::ParseFile — 0 parse errors across all files" },
-    { "name": "Run Pester tests", "status": "Failed", "detail": "Invoke-Pester -Path tools -Output Detailed -PassThru: FailedCount 12, PassedCount 241, TotalCount 253. All 12 failures are in the newly kit-synced Read-DesignState.Tests.ps1/Test-DesignState.Tests.ps1/Test-CIWorkflow.Tests.ps1/Update-DesignProjection.Tests.ps1 — self-referential tests (Split-Path $PSScriptRoot -Parent) that validate the agent-kit's own dogfooded design/state/ per-unit-record mechanism and its verify.yml shape against whatever repo they are copied into. Failing tests: S4.6 (unit/command/track and unit/document/agents-md exist and their closures are complete), the 'Run Pester tests' step carries a GH_TOKEN env test, 'this repository's own table and its own enumeration agree', S5.6 (the real run names a largest closure), S5.12, S7.9, S12.5, S12.6 Describe block (BeforeAll/AfterAll failed), S16.1/S16.2, S16.5, S17.2, S18.6, and Update-DesignProjection's S7.2 real-repo DryRun test. This repository's design/ uses 00-brief/10-design/20-contract/30-slices/90-decisions, not the kit's own design/state/ per-unit-record scheme, so these fail structurally (StateSetAbsent) rather than from a regression — the same class of gap Test-DesignState.ps1 itself reports as exit code 2 when run directly against this repo." },
-    { "name": "Validate Markdown links, terminology, and generated files", "status": "Passed", "detail": "./build/Test-Documentation.ps1 — Human-documentation drift check passed across 18 generated engine page(s), compatibility pointers, and the guide. Documentation checks passed across 123 Markdown file(s)." },
-    { "name": "Build documentation", "status": "DidNotRun", "reason": "docs.ps1 is absent from this checkout — CLAUDE.md states it is installer-generated (Invoke-SetupDocs), not committed, and installing it was out of scope for this change. Needs Docker and an installed docs.ps1; neither precondition was set up this session." },
-    { "name": "Build and verify landing page", "status": "Passed", "detail": "npm --prefix src/engine run build (tsc) exit 0, then npm --prefix site run check: format:check/lint/typecheck/test/test:browser/test:build/test:merge all passed — Test Files 1 passed (1), Tests 11 passed (11) for test:browser; 'Both built HTML entry points contain their required static metadata, and 3 bundle(s) are free of Node-only runtime references'; 'Package-backed merge proven' for test:merge" },
-    { "name": "Merge landing page into documentation build", "status": "DidNotRun", "reason": "npm --prefix site run merge needs a completed docs build in artifacts/docs, which the 'Build documentation' gate above did not produce (docs.ps1 absent)." },
-    { "name": "Test the host", "status": "DidNotRun", "reason": "dotnet test src/host/SubZeroDev.GameEngine.Host.Tests requires NUGET_GITHUB_TOKEN for the sibling-repo NuGet feed; not set in this local environment." },
+    { "name": "Inspect tarball", "status": "Passed", "detail": "tar -tzf the-running-dev-game-engine-0.10.0.tgz: no matches for (^|/)src/, no tsconfig*.json entries, no .test.* build artifacts, package/dist/ present — INSPECT OK" },
+    { "name": "Consumer smoke", "status": "Passed", "detail": "consumer-smoke: cleared node_modules/package-lock.json/dist, then npm run install:engine && npm run build && npm run smoke — all three exited 0, tsc build and smoke.js ran clean" },
+    { "name": "Parse-check PowerShell scripts", "status": "Passed", "detail": "Parsed every *.ps1 (repo-wide) with [System.Management.Automation.Language.Parser]::ParseFile — 0 parse errors: 'All .ps1 files parsed cleanly.'" },
+    { "name": "Run Pester tests", "status": "Passed", "detail": "Invoke-Pester -Path tools -Output Detailed -PassThru — Tests Passed: 235, Failed: 0, Skipped: 18, Inconclusive: 0, NotRun: 0" },
+    { "name": "Validate Markdown links, terminology, and generated files", "status": "Passed", "detail": "./build/Test-Documentation.ps1 — 'Human-documentation drift check passed across 18 generated engine page(s), compatibility pointers, and the guide. Documentation checks passed across 123 Markdown file(s).' Previously failed in this session because design/20-contract.md's §12 event-status edit hadn't been carried through the generation workflow; ConvertTo-HumanDocumentation.ps1 and /make-human-docs were run and the guide stamped, regenerating docs/docs/engine/12-world-graph-kind.md and docs/docs/guide.md, and this re-run now passes." },
+    { "name": "Build documentation", "status": "DidNotRun", "reason": "Runs inside the ghcr.io/the-running-dev/docs-template container image via /template/scripts/docs-build.ps1. docs.ps1 (the local wrapper) is absent from this checkout — CLAUDE.md states it is installer-generated (Invoke-SetupDocs), not committed — and pulling/running the container image directly was not attempted this session; the corresponding CI check ('Verify Documentation Build') on the pull request is where the answer will come from." },
+    { "name": "Build and verify landing page", "status": "Passed", "detail": "npm --prefix src/engine run build (tsc -p tsconfig.build.json) exit 0, then npm --prefix site run check: format:check (prettier — all matched files use Prettier code style), lint (oxlint, clean), typecheck (tsc -b, clean), test (vitest — Test Files 2 passed (2), Tests 15 passed (15)), test:browser (vitest browser-mode via Playwright's managed Chromium — Test Files 1 passed (1), Tests 11 passed (11)), test:build (vite build succeeded, 'Both built HTML entry points contain their required static metadata, and 3 bundle(s) are free of Node-only runtime references'), test:merge ('Package-backed merge proven') — all exited 0" },
+    { "name": "Merge landing page into documentation build", "status": "DidNotRun", "reason": "npm --prefix site run merge needs a completed docs build in artifacts/docs, which the 'Build documentation' gate above did not produce (docs-template container image not available locally)." },
+    { "name": "Test the host", "status": "DidNotRun", "reason": "dotnet test src/host/SubZeroDev.GameEngine.Host.Tests requires NUGET_GITHUB_TOKEN to restore SubZeroDev.Platform.Hosting from the sibling repository's GitHub Packages feed; not set in this local environment. The corresponding CI check ('Build, run, and smoke the image') on the pull request is where the answer will come from." },
     { "name": "Positive route and probe smoke", "status": "DidNotRun", "reason": "Requires a built host image, which itself depends on restoring packages from the sibling-repo NuGet feed (NUGET_GITHUB_TOKEN, unavailable locally)." },
     { "name": "Negative fixture -- corrupted artifact must fail to start", "status": "DidNotRun", "reason": "Same dependency as 'Positive route and probe smoke': requires a built host image, blocked on the missing NUGET_GITHUB_TOKEN." }
   ]

design/20-contract.md

@@ -6249,18 +6249,19 @@ Namespaced `kind.world-graph.*` (05 §9), declared as `Kind.eventNames`:
 | `objective.progressed` / `objective.met` | `debug` / `info` | System 17 | specified, not yet delivered |
 | `failure.progressed` / `failure.triggered` | `debug` / `info` | System 18 | specified, not yet delivered |
 | `scenario.resolved` | `info` | Win or failure, with the `outcome` ids (§8) | specified, not yet delivered |
-| `achievement.unlocked` | `info` | System 19, before alert derivation | specified, not yet delivered |
-| `alert.raised` / `alert.cleared` | `debug` / `trace` | System 19 active-set transition | specified, not yet delivered |
+| `achievement.unlocked` | `info` | System 19, before alert derivation | delivered |
+| `alert.raised` / `alert.cleared` | `debug` / `trace` | System 19 active-set transition | delivered |
 
 > **The status column tracks emit sites, not decisions.** `Kind.eventNames` on the shipped
 > `worldGraphKind` declares the delivered rows; the rest are named here because an event
 > name is a published identifier a sink filters on (05 §9), so it is fixed once, ahead of the
-> emit site, rather than renamed after every host has configured for it. Undelivered rows
-> cluster where the pipeline itself is still incomplete — systems 12, 13 and 19 are no-op
-> stubs and 14 and 16 are partial, recorded with the rest of the tick-system gaps in
-> `90-decisions.md`. Same treatment as `story-graph` §8.4, and safe for the same reason:
-> 05 §2 guarantees dropping every event changes nothing, so a not-yet-emitted event cannot
-> be load-bearing.
+> emit site, rather than renamed after every host has configured for it. As of W85, every
+> tick system this table names is real — `90-decisions.md`'s tick-system register closes
+> with no rows remaining — so the undelivered rows left below belong to pipeline stages
+> (queue membership, guest movement, per-system diagnostics) that were never blocked on a
+> stub, only not yet built. Same treatment as `story-graph` §8.4, and safe for the same
+> reason: 05 §2 guarantees dropping every event changes nothing, so a not-yet-emitted event
+> cannot be load-bearing.
 
 **`guest.path.failed` earns its place.** A resort where guests silently cannot reach a
 building looks identical to one where they do not want to — the failure is invisible in the

design/90-decisions.md

@@ -430,6 +430,8 @@ Reversibility: cheap — documentation only; no code or contract behaviour chang
 **Amended 2026-08-20 by W81.** `construction` (system 12) is real as of this unit and leaves the list below. Re-reading systems 9 and 11 against the source while slicing W81 found the register incomplete rather than wrong: `task-generate` (system 9) and `staff-work` (system 11) were never on this list — both looked complete because every other documented task kind they touch (`clean`, `service`) was wired — but `StaffTaskType` and the `task-assign` comparator have always included `build` and `restock` as options neither system ever produced or applied. **The gap is seven systems, not five.** W81 wired the `build` half of both (a construction-site candidate in system 9, and status marking in system 11, with the effort application and completion itself living in system 12); the `restock` half is still missing and is W82's scope. `buildings` (system 13) and `alerts` (system 19) remain fully no-op; `cleanliness-wear` (14) and `incidents` (16) are unchanged from the description above.
 **Amended again 2026-08-20 by W82.** The `restock` half W81 left open is wired: `task-generate` (system 9) now generates a restock candidate per below-capacity product, `staff-work` (system 11) marks a restocker's task `in_progress` on arrival — mirroring how it already treats `build` — and `buildings` (system 13) applies the assigned restocker's effort to finite inventory, clamped once at each product's capacity, and completes the task. `buildings` is real as of this unit and leaves the list below. `task-generate` and `staff-work` are now fully wired for every documented task kind (`service`, `clean`, `restock`, `build`) and leave this register entirely. **Three systems remain**: `alerts` (19) is fully no-op; `cleanliness-wear` (14) and `incidents` (16) are unchanged from the 2026-08-05 description.
 **Amended again 2026-08-20 by W83.** `cleanliness-wear` (system 14) is real as of this unit for four of its five documented sources: `service` (deferred from system 4), `staff` (deferred from system 11, including cleaning's `onResolve` recovery), `policy` (deferred from system 1), and `litter` (ambient, computed here from unresolved litter-kind incidents) now compose in one sum per building/meter and clamp once, per §9's "systems 1, 4, and 11 explicitly defer building-meter deltas to system 14" rule; wear now moves and a zero-wear open or closed building becomes `broken`. The contract's third ordered slot, `incident`, has no independent mechanism the contract or the source distinguishes from `staff`'s deferred `onResolve` effect (see the `AskUserQuestion` resolution this unit's session recorded) and is left a no-op placeholder in the source's own comment — a future unit that gives it real content should also correct this entry rather than leaving both stale. **Two systems remain**: `alerts` (19) is fully no-op; `incidents` (16) is unchanged from the 2026-08-05 description.
+**Amended again 2026-08-21 by W84.** `incidents` (system 16) is real as of this unit: it resolves due/condition-met occurrences as before, and now also rolls new ones by declared scope (world/zone/building), each definition's own trigger condition, roll chance, and selection weight, blocked by an active or still-cooling-down occurrence of the same definition at the same scope instance, using a per-call `tickRng("incidents")` handle so a split batch stays byte-identical to the whole one (§4.18, §5). `incidents` leaves the list below. **One system remains**: `alerts` (19), fully no-op.
+**Amended again 2026-08-21 by W85, closing this entry.** `alerts` (system 19) is real as of this unit: still-locked achievements are evaluated by definition id against the post-resolution state first, with an unlock writing `unlockedAchievementIds.<id>.exists` under the core `achievement_unlocked` reason; then the three closed alert families — active incident, broken building, and scenario resolved — derive from the current tick's own source set, keyed on published ids only, raising a new alert only for a semantic key no uncleared alert already represents and marking a no-longer-active source's alert cleared. `alerts` leaves the list below. **Zero systems remain — every row this entry ever named is now real, and it needs no further amendment.**
 Reversibility: cheap — documentation only.
 
 ### 2026-08-05 — Known-and-retained implementation gaps: `simulation` end-of-week systems

docs/docs/engine/12-world-graph-kind.md

@@ -1577,18 +1577,19 @@ Namespaced `kind.world-graph.*` (05 §9), declared as `Kind.eventNames`:
 | `objective.progressed` / `objective.met` | `debug` / `info` | System 17 | specified, not yet delivered |
 | `failure.progressed` / `failure.triggered` | `debug` / `info` | System 18 | specified, not yet delivered |
 | `scenario.resolved` | `info` | Win or failure, with the `outcome` ids (§8) | specified, not yet delivered |
-| `achievement.unlocked` | `info` | System 19, before alert derivation | specified, not yet delivered |
-| `alert.raised` / `alert.cleared` | `debug` / `trace` | System 19 active-set transition | specified, not yet delivered |
+| `achievement.unlocked` | `info` | System 19, before alert derivation | delivered |
+| `alert.raised` / `alert.cleared` | `debug` / `trace` | System 19 active-set transition | delivered |
 
 > **The status column tracks emit sites, not decisions.** `Kind.eventNames` on the shipped
 > `worldGraphKind` declares the delivered rows; the rest are named here because an event
 > name is a published identifier a sink filters on (05 §9), so it is fixed once, ahead of the
-> emit site, rather than renamed after every host has configured for it. Undelivered rows
-> cluster where the pipeline itself is still incomplete — systems 12, 13 and 19 are no-op
-> stubs and 14 and 16 are partial, recorded with the rest of the tick-system gaps in
-> `90-decisions.md`. Same treatment as `story-graph` §8.4, and safe for the same reason:
-> 05 §2 guarantees dropping every event changes nothing, so a not-yet-emitted event cannot
-> be load-bearing.
+> emit site, rather than renamed after every host has configured for it. As of W85, every
+> tick system this table names is real — `90-decisions.md`'s tick-system register closes
+> with no rows remaining — so the undelivered rows left below belong to pipeline stages
+> (queue membership, guest movement, per-system diagnostics) that were never blocked on a
+> stub, only not yet built. Same treatment as `story-graph` §8.4, and safe for the same
+> reason: 05 §2 guarantees dropping every event changes nothing, so a not-yet-emitted event
+> cannot be load-bearing.
 
 **`guest.path.failed` earns its place.** A resort where guests silently cannot reach a
 building looks identical to one where they do not want to — the failure is invisible in the