Instructions For Codex

Instructions for Codex

Claude-owned handoff (2026-06-02). A single, current, prioritized action queue for the wiki-quality + audit work that falls in Codex's lane (atlas/nav/structure/status pages). Detail and evidence for each item live in Wiki quality audit (sections A/B/C + Round 2). Claude-lane items are already done. Work top-to-bottom; check items off in place. Acceptance = the stated condition is true and the page still passes the link gate.

Already done (no action)

• ✅ Nav: all 5 new Claude pages are wired into _Sidebar.md (Wiki-Quality-Audit, Variable-And-Naming-Conventions, Public-Variable-Channel-Index, Modules-Atlas, Pending-Owner-Decisions).
• ✅ Dedup routing (audit section A): construction / supply-cooldown / generated-mission / lifecycle / victory / BattlEye duplications resolved.
• ✅ Accuracy C1: Networking MASH row now cites DR-34 (was the stale "server relay live"/DR-3).

P0 — Accuracy (a reader is currently being misinformed)

1. Fix the UI-atlas finding mislabels (Client UI systems atlas line ~254 "Buy-gear authority" row, and the same in Client UI HUD and menus). The row lists "DR-16, DR-17, DR-24" under gear/template — but DR-16 = structure-sale client-authority (not gear), DR-17 = duplicate IDD 23000, DR-24 = dead RscMenu_Upgrade. And DR-25a = duplicate title IDD 10200, DR-25b = malformed soundPush[] (not "EASA/service authority" — that's DR-28). Acceptance: each DR is described as its actual finding; the gear/EASA authority risk links DR-28.
2. De-hedge the MASH status in SQF atlas — replace "MASH marker status requires careful source verification" with a definitive DR-34 cite (dead both ends). Acceptance: no "requires verification" hedge for MASH.
3. Mark the compile counts point-in-time in SQF atlas — the "659/452/207" numbers need a timestamp + regeneration command + DR-5 cite (DR-5 warned they go stale). Acceptance: counts labelled point-in-time with how to regenerate.

P1 — Surface findings where developers actually look (orphaned cross-links)

4. DR-44 (wfbe_supply_temp_<side> forgery) → add to Economy synthesis table, Networking direct-channel section, Server runtime atlas.
5. DR-20 (HQ-killed N-fold score exploit) → add to Construction atlas, Gameplay atlas, Server runtime atlas.
6. DR-45 (town-AI deletes player-occupied vehicles) → cross-link from Town-AI-Vehicle-Despawn-Safety and AI/headless.
7. Cite DR-40 by number in WASP overlay and DR-19 in Server runtime atlas.
8. Add the missing DR cross-links called out in audit section C2 (Gameplay atlas: DR-6/14/11/22/23/15; UI atlases: DR-16/17/24/25; AI/headless: DR-21/42; Construction atlas: DR-6; Mission entrypoints: DR-37/43a).

P1 — Reconcile "current work" status pages (they're stale)

9. Coordination board: the "Active Lanes" + "Roles" tables are stale — sub-agent lanes (Faraday/Mencius/Hilbert/Cicero/Curie/Meitner) shown "Active" were harvested/closed in Wave F; victory-endgame-runtime-atlas shown "Active" is integrated; the Roles line says "Claude's latest reviews are DR-11..DR-15" but Claude is at DR-45 and in collaboration-follow mode. Acceptance: tables reflect current state (closed lanes marked closed; Claude at DR-45).
10. Progress dashboard: the "At A Glance" Claude row predates Phase-1 completion + the collaboration-follow operating mode; update to reflect that the coverage ledger's Map/Perf/JIP-HC columns are complete and only Auth/PV owner-decisions remain.
11. _Sidebar.md: Headless-Delegation-And-Failover-Playbook is listed twice (Gameplay + Ops) — de-duplicate.

P2 — Structure (page merges, audit section B)

12. Merge Hardening roadmap + Server authority map (~70% overlap) — Migration-Map's Authority Principles + Handler Validation Checklist as preamble, Roadmap's work packages as body.
13. Reduce Client UI HUD and menus to a quick-ref/redirect into Client UI systems atlas (it is a strict subset).
14. Trim Mission entrypoints ↔ Lifecycle wait-chain overlap (~50%): keep the description.ext include graph + mission-object init layer on entrypoints, the boot-timeline/flag-dependency graph on wait-chain, cross-link.

P3 — Thin citations (lower priority)

15. Add representative path:line anchors to Core systems index, Architecture overview, Content structure and maps.

Do NOT change (audit false positives — verified at source)

• Public variable channel index PVF list ranges :8-20 (server) / :23-37 (client) are correct.
• DR-15's _side = _this at Server_AssignNewCommander.sqf:3 is correct. (An audit pass miscounted blank lines; leave both as-is.)

Not Codex's lane — informational

• The economy/forgery owner decision (server-authority vs BattlEye, two surfaces) and the coverage-gap next-review queue (Server/AI respawn+orders, cleaners Perf, config data model, PR#1 delta) are in Pending owner decisions and Deep-review findings Round 36 — for the owner / the next review phase, not a Codex doc task.

New findings — document + cross-link (2026-06-02, Claude)

A standalone "20 improvements" review (a shareable PDF for the owner) source-surfaced one genuinely new finding and several maintainability leads:

16. DR-46 — SEND_MESSAGE second call compile RCE (Deep-review findings). Cross-link it from Networking, Public variable channel index (the SEND_MESSAGE row) and SQF atlas (the call-compile inventory). Also correct DR-1's "no other network-data compile sites" note (≈ line 159) to acknowledge the SEND_MESSAGE direct-PV surface — the repo has two network-data compile RCE surfaces, not one. Acceptance: DR-46 is reachable from the three atlas/index pages and DR-1 no longer implies a single compile site. Fix shape: agent-hardening-backlog.jsonl#send-message-call-compile-rce.
17. Maintainability leads — verify, then document only if not already owned by Variable and naming conventions / SQF atlas: Init_Common.sqf double-compiles ~15 fns under both GetX and WFBE_CO_FNC_GetX (plus ~22 un-prefixed in Init_Server.sqf, four GetClosestEntity{,2,3,4} variants); the Client_Support*, Construction_*Site and Loadout_* copy-paste families; ~12 hardcoded English hints bypassing stringtable.xml. Acceptance: each lead is either cross-referenced to its existing owner page or recorded as a new maintainability note.
18. DR-30 correction (BattlEye) — drop remoteexec.txt from the missing-filters list (it is Arma 3-only; A2 OA has no remoteExec), name the A2 OA filter set instead, and reframe BattlEye as contingent, local-filter defense-in-depth (BE's online support for A2 is defunct; Client/FSM/updateclient.sqf:153-162 kickAFK is the evidence filters are processed locally). Propagate to External integrations (BattlEye section) and any "missing filters" list; DR-30's "no filters shipped" substance stands. Acceptance: no page lists remoteexec.txt as a missing A2 filter, and BattlEye is framed as defense-in-depth behind server-authority. Detail in Deep-review findings "DR-30 correction".

Batch-4 consistency findings — document / correct (2026-06-02, Claude)

Read-only wiki↔source audit of 7 playbook/index pages (~190 claims, 29 confirmed, all re-verified at source). Attack-wave authority playbook passed clean (0/8). Ordered by severity:

19. Supply-mission authority cleanup playbook — HIGH: documents code that does not exist in current source. It reads like a draft "PR #1" written up as the shipped state. Nonexistent in source: (a) a Killed event handler on the supply vehicle — 0 grep hits for addEventHandler/Killed anywhere under Server|Client/Module/supplyMission/ (the page's Step 4 + findings-table both describe one at :10-25 and warn it "can stack"); (b) a SupplyByHeli object var — 0 grep hits across the whole mission (Step 3 + Step 7 claim it is stamped + read); (c) a heavy-heli cash-run + commander-team-funds completion branch (Step 7) — supplyMissionCompleted.sqf (36 ln) reads only SupplyAmount/SupplyFromTown and calls only ChangeSideSupply:26 (supply only, no funds path); (d) recommended-patch item 4 cites constants WFBE_C_SUPPLY_TRUCK_TYPES/WFBE_C_SUPPLY_HELI_TYPES_T2/_T3 that don't exist — truck types are hardcoded inline at supplyMissionStart.sqf:18 and Skill_Apply.sqf:72. Also the status table marks the scan "Source/Vanilla patched", but supplyMissionStarted.sqf:28 is still nearestObjects [..., [], 80] (empty filter — DR-39 not applied), and the page contradicts itself (Step 5 correctly shows []). Acceptance: the page is either clearly labelled a proposed/not-yet-applied PR, or each step is corrected to current source; the scan status is no longer "patched"; nonexistent vars/constants are removed.

20. ICBM authority playbook — MEDIUM: a normative line reads as a factual status. "Client Tactical-menu checks are affordances; server-side validation remains the authority boundary" implies server validation exists. It does not: Server_HandleSpecial.sqf:97-112 takes _side/_base/_target/_playerTeam from the request, logs, then only if (isNull _target || !alive _target) exitWith {} and spawns NukeDammage — no commander/upgrade/funds check (fully client-authoritative, DR-27). Acceptance: reword to normative ("should remain") and link DR-27 inline.

21. Function and module index — MEDIUM + LOW. (MED) Common/Module/Reaktiv is listed "Live" but Reaktiv_Init.sqf has zero callers outside its own folder (never compiled in Init_Common.sqf; contrast IRS at Init_Common.sqf:320) → dead/unreachable. (LOW family-summary drift) Common_Get* lists "depots" but the depot getter is Client_GetClosestDepot (Client family, Init_Client.sqf:102); Common_Set* lists "waypoints" (no waypoint setter exists; closest is SetTeamMovePos); Common_Handle* lists "damage" (no Common_HandleDamage); Server_AI_Com_Upgrade is written WFBE_C_UPGRADES_%SIDE_AI_ORDER but source uses Format ["WFBE_C_UPGRADES_%1_AI_ORDER", _side] (a grep for %SIDE finds nothing). Acceptance: Reaktiv marked dead; the four family/name descriptions corrected.

22. PVF dispatch implementation playbook — MEDIUM + LOW. (MED) the client-command enumeration lists 9 names as if complete, but _clientCommandPV (Init_PublicVariables.sqf:25-41) has 14 active entries — missing AwardBountyPlayer, ChangeScore, SetTask, SetMHQLock, Available. The fix code uses forEach _clientCommandPV so the patch is fine, but a reader hand-building an allowlist from the prose would drop 5 channels. (LOW) off-by-one ranges: compile loops :43-50→:43-51; Common_SendToServer/…Optimized :12-17→:12-18; …SendToClient :13-20→:13-21; …SendToClients :12-18→:12-19; and supplyMissionCompletedMessage.sqf RequestChangeScore cited :32 but it's :22 (file is 24 ln). The call compile RCE mechanism + fix-snippet SQF (A2-OA-valid) are correct. Acceptance: client list enumerates all 14 (or hedges like the server list); ranges corrected.

23. Economy authority first cut — MEDIUM. Same SupplyByHeli ghost var (supplyMissionStart.sqf:34-46 cited; var doesn't exist; SupplyFromTown is :20, not in range). "Rewards funds/supply at :31-40" — only supply is rewarded (ChangeSideSupply:26); no funds path in supplyMissionCompleted.sqf (0 grep for ChangePlayerFunds/ChangeTeamFunds). Acceptance: drop SupplyByHeli; correct the reward description to supply-only.

24. Variable and naming conventions — LOW. WFBE_CO_FNC_LogContent is presented as compiled in Init_Common.sqf but its definition site is initJIPCompatible.sqf:37; ≥4 other WFBE_CO_FNC_* are compiled in Init_Server.sqf (InitAFKkickHandler:63, LogGameEnd:64/89, DelegateAITownHeadless:99, DelegateAIStaticDefenceHeadless:100) — the prefix isn't exclusively an Init_Common family. wfbe_funds is set on group objects (Init_Server.sqf:474, Common_ChangeTeamFunds.sqf:8), not the logic/side/town/building objects the row implies. The wfbe_* row's lead example "queu" (Client_BuildUnit.sqf:172) doesn't carry the wfbe_ prefix. Acceptance: soften "compiled in Init_Common" to "mostly", note the group-object case, swap the queu example.

25. Batch-2 recovery → owning pages (5 still-live items lost to a shared-worktree clobber of commit 89f848f; all re-verified at source today). Routing to the owning pages rather than re-adding a section to the findings page (your call on that page's shape):

    • Modules atlas: (:18) the incomingMissile EH at Client_BuildUnit.sqf:342 spawns HandleATMissiles (generic AT), not IRS — only :356 spawns WFBE_CO_MOD_IRS_OnIncomingMissile. (:21) the aircraft CM-removal block is gated if !(WF_A2_Vanilla) at Client_BuildUnit.sqf:275 → it runs on OA, so "Vanilla-only by design" beside the removal ref is misleading (the CM module is vanilla-gated at Init_Client.sqf:590; the removal is the non-vanilla path). (:41) Soldier skill is ceil(1.5 × …) (Skill_Init.sqf:49) — 1.5×, not "double team size". (:47) CIPHER_Sort is an ExecVM'd script (Labels_Upgrades.sqf:127 → Common\Module\CIPHER\CIPHER_Sort.sqf), not a compiled utility fn; the compiled CIPHER fns are CIPHER_SortArray/CIPHER_SortArrayIndex (CIPHER_Init.sqf:58,94).
    • Economy towns and supply (:30): "player delivery funds use WFBE_C_PLAYERS_SUPPLY_TRUCKS_DELIVERY_FUNDS_COEF" — that constant (Init_CommonConstants.sqf:268, =4) has zero consumers; the award passes raw _supplyAmount to ChangePlayerFunds (supplyMissionCompletedMessage.sqf:13). Dead constant.

Continue Reading

Detail: Wiki quality audit | Findings: Deep-review findings | Decisions: Pending owner decisions