Eve library

V2-002 — Manual live-ops lifecycle reconciliation evidence

Evidence

Revision 1

On this page
  • Owner: Codex implementation agent
  • Accountable product: V2
  • Priority: P0 flagship-evidence support
  • Linked task: V2-002, bring manual MOTD and patch-note lifecycle into the common ledger
  • Shared inventory implementation source: 968c65790ff87e457c8cb7ccc8aff38a6ad38cae
  • Shared inventory remote-integrated source: f0665ab54fdf1f62bea260ba51f023bc45c2f52e
  • Current source-digest refresh: 2d29a2fc3b3d12840ee308eb4d99cef9daec70f4
  • Evidence captured: 2026-07-24

Reconciled boundary#

The canonical COV-004 inventory places V2's three human-authored message-of-the-day definitions and its structured patch-note release in the same manual-content lifecycle and lineage ledger used by the rest of the portfolio. Each row has a stable row identity, exact source locator and source-file SHA-256, lineage reference, locale observation, reviewed accessibility requirements, lifecycle state, quantities, explicit gap kinds, and assigned migration owner.

The MOTD rows are stale because their effective windows expired before the inventory date. The patch-note row remains review; neither source presence nor the in-game service contract is represented as production placement. This preserves the distinction between manual lifecycle coverage and the live placement/runtime evidence required elsewhere in V2-002.

The V2 source services remain the authoritative authored records:

  • @v2/message-of-the-day-service validates event, downtime, and patch-note announcements, active windows, targeting, category metadata, and per-account dismissal.
  • @v2/patch-notes-service validates structured balance, bug-fix, and new-content entries and produces the in-game grouped display payload.
  • COV-004 extracts the four manual records from those sources and fails on source-digest drift, lifecycle weakening, count drift, missing lineage, or unowned gaps.

Verification#

Command Result
pnpm --dir apps/v2/message-of-the-day-service test PASS — 1 file, 4 tests
pnpm --dir apps/v2/message-of-the-day-service typecheck PASS
pnpm --dir apps/v2/message-of-the-day-service lint PASS
pnpm --dir apps/v2/message-of-the-day-service build PASS
pnpm --dir apps/v2/patch-notes-service test PASS — 1 file, 3 tests
pnpm --dir apps/v2/patch-notes-service typecheck PASS
pnpm --dir apps/v2/patch-notes-service lint PASS
pnpm --dir apps/v2/patch-notes-service build PASS
python3 V2/ue/Tools/check-v2-message-of-the-day.py PASS
python3 V2/ue/Tools/check-v2-patch-notes.py PASS
pnpm verify:content-coverage:inventory PASS — 10/10 tests, deterministic output current, script/contracts typechecks and focused ESLint green
pnpm exec tsx scripts/content-coverage/generate-content-inventory.ts --check after remote integration PASS — 77 rows, 14 manual

Artifact hashes#

Artifact SHA-256
content-coverage/content-inventory.v1.json cec5dff73c23e28c51b41377f072e9e81c367415d1152be966456f48a373c08d
apps/v2/message-of-the-day-service/src/message-of-the-day-service.ts 2820b2a1f7d5ab448375ac76b99e295443bec9c0e23b664a934408ae01a4a983
apps/v2/patch-notes-service/src/patch-notes-service.ts b4322391c29678f8dc55dd530f55acd583892e6167df3dd31e88059f12cb8826

Honest boundary#

This evidence closes only the V2-002 manual-content lifecycle-ledger item. It does not claim that any MOTD or patch note is currently placed in production, localized beyond its recorded source locale, approved for a new release, or observed by a client. It also does not satisfy the real-match commentary, localized media, consented voice, face animation, signed provenance, Unreal cook/playback, operator intervention, latency, or cohort-preference bullets, which remain open.

Flagship execution evidence gate#

Evidence-gate source commit: b238631d1a2af477815a7ebfb6f2a36794b6bde7.

The five remaining flagship requirements now map one-to-one to a fail-closed fresh-run contract:

  • realMatchCommentary requires one real match, distinct live and post-match provider receipts, an authoritative snapshot, a fact-proof bundle, and all claims grounded.
  • localizedMediaRuntime requires localized captions, consented voice, face animation, signed provenance, immutable runtime media, and observed playback bound to V2/ue/V2.uproject on ue55-linux.
  • ueCookAndAssets requires an executed RunUAT BuildCookRun for target Linux, run as ueagent, bound to the exact V2 project, with Isis asset IDs and a hashed cooked archive.
  • operatorIntervention requires pause, edit, replace, approve, and rollback through the actual UI, each with actor, timestamp, receipt, and a passing measured latency bound.
  • cohortEvaluation requires a policy-sized production-representative cohort and signed measurements for naturalness, factuality, usefulness, latency, and preference.

The executable contract is scripts/v2/verify-flagship-evidence.mjs. The retained input and readiness report are V2-002-flagship-evidence.json and V2-002-flagship-readiness.json. The current report is blocked on all five checks because no coherent live match, UE runtime/cook, UI intervention, or cohort package is retained. The 2026-07-27 missing-Unreal observation is superseded by evidence/v1-v9/V2-unreal-host-preflight.json: UE 5.5.4, RunUAT.sh, ueagent, and the required EOS editor binary are present, but the retained V2 editor build ID does not match the current engine build ID. A resource-safe V2Editor rebuild is required before qualified automation or cook; that rebuild still cannot substitute for Isis asset IDs, real match media, runtime playback, operator receipts, or cohort evidence. No V2-002 checkbox is closed by this preflight or gate.

Verification on 2026-07-27 passed Prettier, ESLint, node --check, and all 17 V2 evidence tests (8 flagship and 9 V2-001 live-execution cases). The default CLI exits nonzero for the current evidence; --report reproduces the retained readiness JSON exactly after formatting. The 2026-08-02 preflight suite adds 5/5 passing host-state tests, and the retained real-host record verifies byte-for-byte against the current prerequisites. The shared preflight refactor adds a sixth V2 case proving that absent project modules are rebuild-required on a capable host rather than host-unavailable.

sh
node scripts/v2/verify-flagship-evidence.mjs \
  --evidence=evidence/v1-v9/V2-002-flagship-evidence.json \
  --report