Forever simulations

Level 60 · five-minute single target · 5,000 iterations per race/build · full role-specific Tier 1 bonuses.

Rows marked Tank include incoming attacks and modeled healing; their rage, threat and survival are scenario-dependent. See tank metrics.

All equipment is hypothetical level-65 modeled gear; real enchants remain. Hit comes from items, enchants, talents and racials, with no paid conversion. Ordinary MP5; no world or raid campfire buffs. These are tested profiles, not proven global optima or measured class balance.

Questions & coverage

137/137 questions reviewed; 58 new coverage gaps. Sources, remaining tests and implementation work are separated below.

Latest follow-up (2026-10-16): October 16 follow-up: Shaman imbues and Rogue poisons in their own enchant slot beside stones and oils; real idols, librams and totems with a quarter of the modeled relic stats, chosen by the gear search; Rake Attack Power scaling (the September 30 deep dive, fitted to beta logs) and Rake in the Feral rotation; Cat and Bear proc rates at the form speed; bear-weaving screened (a loss); Frost Cone of Cold at five Winter’s Chill stacks, Shadow full Mind Flay channels, Smite Shadow Word: Pain, Enhancement Shadow Oil. Every DPS profile re-searched with two-seed confirmation; all 1,005 matrix scenarios and 68 tank checks replayed. Latest follow-up report.

Original adjudication baseline: 2026-09-27, 35cab052456a320358e998c12416687d69cae142. That original review reconciled questions without changing mechanics or benchmark results; later implementation and benchmark updates are included in this current register and linked above. A source-backed correction is not necessarily implemented: use each record’s current status. “Beyond level 30” means outside announced beta access; move it earlier if access expands.

Review accounting and evidence

Status	Before	Original entries now	Including additions
Needs evidence	84	38	45
Implementation gap	27	3	3
Model assumption	5	3	6
Accepted working answer	0	13	24
Resolved	20	50	75
Outside current scenarios	1	42	42

Accepted working answers are qualified, not verified server facts. Confidence applies to the stated finding. Review summary · Adjudications · Source ledger

Prioritized test and implementation plan

High priority first. A question may have partial checks at several stages. Accepted-answer corroboration is optional and remains within its record.

Current beta · level 30 · 28 questions

An experiment within the current level-30 beta, subject to actual access, learned ranks and prerequisites. Sixteen Legacy Points and Dalaran expand opportunities, not the class talent budget or level-60 access.

CORE-001 — Boss attack tables High
Requires: Known skill/hit, an equal-level and a +3-level target, recorded facing.
Observe / verify: Count white/special/ranged outcomes separately; compare queued and unqueued attacks. This supplies a low-level table, not a level-63 certification.
CORE-002 — Energy regeneration and haste High
Requires: Two known general-haste totals, identical resource talents, uncapped recovery windows and precise resource timestamps.
Observe / verify: Repeat matched recovery cycles from low Energy to below the cap, with no attacks or refunds. Aggregate endpoint uncertainty; a single rounded 1% window is insufficient. Run an attack-speed-only control separately.
CORE-004 — Spell GCD versus physical GCD High
Requires: Known casting-speed modifier, one magical instant and one physical 1.5-second action; client cooldown timestamps.
Observe / verify: Compare GCD duration independently of cast time. Only test the floor when enough haste is actually available; a 1% modifier cannot establish it.
DATA-001 — Server downranking penalties High
Requires: Two learned ranks of a caster damage spell, controlled SP-only gear changes, fixed target level/resistance and no changing damage talents.
Observe / verify: Compare the damage slope per SP at each rank, not their raw hit sizes. Record exact player level and exclude crits; repeat after leveling if possible.
DATA-001 — Server downranking penalties High
Requires: Level 30 Hunter with Improved Wing Clip and independently accessible lower/higher rank casts, fixed target eligibility.
Observe / verify: Estimate landed-hit proc probability by rank/player level, holding talent and target fixed; current highest-rank-only registration cannot answer this.
DRU-003 — Form weapon damage and flat imbues High
Requires: Bear and level-20 Cat, matched-stat weapons with substantially different DPS, plus an obtainable flat-damage enchant/imbue.
Observe / verify: Compare form attack damage at fixed AP; test the flat effect independently from the weapon swap.
DRU-005 — Bear threat multipliers High
Requires: Bear, white attacks/Swipe, precise threat readings, known damage and no Thorns/proc interference.
Observe / verify: First determine the form multiplier from ordinary hits, then Swipe's additional threat; include multiple enemies without duplicating totals.
DRU-011 — Proc enchants in animal forms High
Requires: Only if an enchanted weapon can actually be supplied in the beta; an obtainable enchant is not established by a low character-level requirement.
Observe / verify: Compare isolated caster-form and Cat/Bear attacks with proc timestamps and equipped/form weapon speeds.
DRU-012 — Bear outgoing and incoming rage formulas High
Requires: Bear, uncapped rage, no Maul/Enrage/Primal Fury and matched attack speed while damage varies.
Observe / verify: Separate ordinary hits, crits and avoided swings. For incoming rage stop autos and vary armor independently from maximum health.
DRU-017 — Rake Attack Power scaling High
Requires: A level-30 Cat with Rake; a way to change Attack Power by a known amount (Battle Shout, a Blessing of Might or an AP item).
Observe / verify: Rake tick damage (non-crit) with and without the Attack Power change, on a target without bleed modifiers.
HUN-001 — Pet owner-stat inheritance High
Requires: Hunter or Warlock pet, one isolated owner-stat change and pet API/damage observations.
Observe / verify: Prioritize demon AP/SP/mana and pet update timing; keep owner and external pet buffs distinct. Do not repeat every accepted Hunter coefficient merely because a server script is hidden.
HUN-002 — Pet per-hit damage versus speed High
Requires: Two same-family/same-level pets of different speeds and two owner AP totals.
Observe / verify: Compare the change in pet damage per hit across the AP delta for each speed; control happiness and training. This separates an AP coefficient from the already observed base-speed difference.
PAL-006 — Seal of Fury shield replacement and mana High
Requires: Level 20+ Paladin with Improved Seal of Fury, shield, learned ranks, uncapped mana and known incoming attacker levels.
Observe / verify: Measure successive procs, rank switching, partial/full depletion, replacement/expiry and capped attacker-level returns; distinguish generated capacity, absorbed damage and mana.
PAL-008 — Vindication's target AP reduction High
Requires: Paladin with 1/3 to 3/3 Vindication (10 Retribution points first) and matched enemy swings.
Observe / verify: Validate the target aura and damage reduction at each rank, without confusing owner AP growth with enemy AP loss.
WAR-006 — Flurry amount and charge events High
Requires: Level-30 Warrior with 26 Fury points including Death Wish and 1/5 Flurry (for example Fury 25053005040010100, no Arms or Protection points); advanced combat logging.
Observe / verify: Auto-attack a mob for three minutes with one weapon, then dual wielding, then with Heroic Strike queued every swing. Count white swings between each Flurry gain and fade, and note whether the buff icon shows 3 charges. About 60 Flurry procs per setup.
WAR-008 — Shield attacks and weapon procs High
Requires: Shield Bash, shield and an actually obtainable known weapon proc; autos stopped.
Observe / verify: Count proc events following isolated Bashes. This is a partial test, not certification of Shield Slam or Windfury.
WL-009 — Demonic Knowledge ranked values High
Requires: Level-30 Warlock with 1/3, 2/3 and 3/3 Demonic Knowledge (20 Demonology points first; 26 points are available), fixed owner gear and an active pet.
Observe / verify: Measure owner and pet spell power at each rank, despawning and resummoning after each change.
CORE-009 — Eureka cast and periodic bonuses Medium
Requires: Gnome Mage/Warlock/Priest with a low-rank DoT or channel; exact build and buff-stack timestamps.
Observe / verify: Repeat those three cases in game. Do not infer intended behavior from an open bug report or conflate the old mana-discount amount with current values.
DRU-018 — On-hit effects at the Cat and Bear form attack speed Medium
Requires: A Cat with a slow (3.0+ speed) weapon carrying a PPM effect, such as Crusader, Fiery Weapon or a chance-on-hit weapon.
Observe / verify: Procs per landed white hit and per landed Shred or Claw over a few hundred of each; compare with PPM x 1.0 / 60 and PPM x weapon speed / 60.
HUN-003 — Pet external buffs and action cadence Medium
Requires: Only if pet cadence sensitivity proves material; a trained damage ability and a resource trace.
Observe / verify: Compare ability use when resource-limited versus resource-capped. Do not mix this with owner inheritance or ordinary spell cooldowns.
HUN-012 — Savage Strikes article/client version conflict Medium
Requires: Savage Strikes access and matched talent-on/off attacks at fixed AP, weapon and target.
Observe / verify: Separate hit-size changes from crit-frequency changes, and record build/ability IDs.
HUN-016 — Fire trap Attack Power scaling Medium
Requires: Hunter with Immolation Trap and two gear sets differing only in melee or only in ranged Attack Power.
Observe / verify: Total trap damage per set, about 20 traps each.
PAL-004 — Holy melee partial resistance Medium
Requires: Holy Strike and seal attacks against targets of known equal and higher levels; fixed damage inputs.
Observe / verify: Inspect explicit resisted amounts and damage clusters separately from miss/dodge outcomes. A small sample with no resist is not proof of immunity.
PAL-010 — Retribution Aura provider scaling Medium
Requires: Paladin level 16+ or Druid Thorns, distinct caster/recipient, fixed enemy and two caster-SP controls.
Observe / verify: Measure base/SP and change caster gear after application; hold recipient gear fixed. Check party/range and damage/threat ownership separately.
PRI-003 — Inner Focus periodic crit behavior Medium
Requires: Level-20 Priest with Inner Focus and a DoT in its crit mask.
Observe / verify: Compare three arms: no aura, aura activated after the DoT and held, and aura consumed by applying the DoT. Record enough ticks and the exact build; separate live observation from intended design.
ROG-005 — Hack and Slash hand scope and ICD Medium
Requires: Level-30 Rogue with 5/5 Hack and Slash (20 Combat points first; 26 points are available) and matched dagger/fist, mace and sword pairs.
Observe / verify: Swap MH/OH placement. Compare each hand's crit and armor-sensitive damage, plus a non-weapon spell control; separate a displayed character-sheet bonus from hand-specific outcomes.
SCEN-005 — General absorb-pool and shielding threat Medium
Requires: Priest with Power Word: Shield, a cooperating player and a controlled hostile target with threat observation.
Observe / verify: Separate threat on shield application, actual absorption and unused expiry; do not attribute incidental healing or damage threat to the shield.
WAR-010 — Weaponmaster hand-specific armor ignore Medium
Requires: Level-30 Warrior with 1/5 to 5/5 Weaponmaster (20 Arms points first; 26 points are available), comparable mace and non-mace weapons, matched AP and controlled target armor.
Observe / verify: Move the mace between hands and measure main-hand damage changes; separate the weapon's own damage change and sword extra attacks. Repeat at 1/5 and 5/5 to confirm rank scaling.
WL-006 — Improved Imp's extra dummy effect Medium
Requires: A live Imp, ranks 0/1/2/3 of Improved Imp, fixed owner/pet haste and mana.
Observe / verify: Record Firebolt cast start/end and the interval before the next cast. Distinguish a shorter cast from removal of the engine's artificial 200-ms recast delay.

Level-30 beta · 9 questions

Requires reachable level-30 abilities, talents and items. Check training, prerequisite rows and Legacy access; do not invent a new 51-point allocation.

DATA-001 — Server downranking penalties High
Requires: The same spell/rank and controls at level 30.
Observe / verify: Check whether its SP slope changes as the player outlevels the rank. This extends the low-level test, not the level-60 conclusion.
DRU-005 — Bear threat multipliers High
Requires: Primal Bite talent, available from 25.
Observe / verify: Isolate direct Bite damage and threat with the previously measured form multiplier.
WAR-001 — Level-60 outgoing rage normalization High
Requires: A known weapon retained across levels and no other rage gains.
Observe / verify: Optional extrapolation checkpoint: compare uncapped rage per landed swing before and after leveling. This does not certify level 60.
WAR-002 — Off-hand and extra-swing rage High
Requires: A sourced extra-attack effect actually available by level 30, and separate MH/OH events.
Observe / verify: Isolate extra swings from regular autos and energizes. If no suitable effect is obtainable, defer that subtest rather than substituting ordinary swings.
WL-001 — Demonic Brand power source and threat High
Requires: Level 25+ Brand, Imp and a melee demon, controlled owner school power and pet modifiers.
Observe / verify: Separate normal pet attacks from Brand events. Vary owner power, record Brand crits and threat, and test a second target without a Brand.
DRU-008 — Nature's Grace GCD stacking and timing Medium
Requires: Nature's Grace, cooldown timestamps and controllable Wrath travel distance.
Observe / verify: Compare GCDs with the aura and an isolated cast-speed control; compare aura activation with cast completion and projectile impact.
MAG-004 — Hot Streak interruption and new-charge ordering Medium
Requires: Level 25+ Hot Streak/Pyro build and cast/impact/aura logs.
Observe / verify: Start accelerated Pyro while an earlier Fire projectile is in flight; compare aura stacks at impact and completion. Separately cancel Pyro if interruption support is being evaluated.
ROG-001 — Mutilate off-hand flat damage and Cold Blood Medium
Requires: Level 30 Mutilate/Cold Blood allocation, two known daggers, fixed AP and target armor.
Observe / verify: Record both child hit/crit outcomes with and without Cold Blood. Swap hand placements and vary AP independently to distinguish flat bonus from weapon contribution.
ROG-002 — Rogue finisher and poison scaling Medium
Requires: Deadly Poison access and a controlled AP delta.
Observe / verify: Measure one-stack ticks and then five-stack ticks to locate the 0.0045 ratio. Audit finishers in separate AP/CP controls rather than mixing poison proc counts with their damage.

Beyond level 30 / launch · 26 questions

Requires access beyond the announced level-30 beta or max-level content. Reclassify if later beta access is announced.

CORE-001 — Boss attack tables High
Requires: Level-60 characters and level-63 targets on a build without the known glancing defect.
Observe / verify: Repeat outcome and crit-suppression counts; compare damage of glances with ordinary hits at fixed AP.
CORE-007 — Judgement of Wisdom trigger rules High
Requires: A level-38+ Paladin with Seal/Judgement of Wisdom, uncapped mana recipients and isolated event sources.
Observe / verify: Count mana returns per white hit, avoided white, special, direct spell, periodic tick and triggered effect. Compare event intervals to distinguish chance from ICD; the level-30 cap does not grant this seal.
CORE-011 — Base attributes and class conversions High
Requires: Naked level-60 characters with racial/talent bonuses accounted for, plus isolated Agility/Intellect gear deltas.
Observe / verify: Record base attributes, health/mana and crit slopes for each class; race pairs isolate racial offsets.
DATA-001 — Server downranking penalties High
Requires: Level 60 and the actual downranked spells used by the retained APLs.
Observe / verify: Repeat the two-SP slope comparison for low and maximum ranks; separate learned-level base growth from the coefficient.
DRU-003 — Form weapon damage and flat imbues High
Requires: Level-60 forms and high-level weapon effects if low-level results imply scaling.
Observe / verify: Verify the actual level-60 base and any effect-specific exceptions before changing synthetic-weapon valuation.
DRU-005 — Bear threat multipliers High
Requires: Lacerate from 42 and eventually level-60 ranks.
Observe / verify: Separate application and bleed threat at each stack; distinguish flat threat from damage multipliers.
DRU-006 — Lacerate direct and periodic formula High
Requires: Level-42+ Bear, stable AP, no other bleeds, isolated first through fifth applications.
Observe / verify: Determine whether the first application deals direct damage and whether subsequent hits use the pre-existing or new stack count; separately measure tick damage and crits.
DRU-011 — Proc enchants in animal forms High
Requires: Crusader access if unavailable in either beta cap.
Observe / verify: Repeat the form comparison with enough eligible attacks to distinguish no procs from a low rate. Move earlier if vendors or copied items make the enchant obtainable.
DRU-012 — Bear outgoing and incoming rage formulas High
Requires: Level-60 Bear and known outgoing/incoming attacks.
Observe / verify: Check that the low-level formula still fits; test queued Maul separately because it replaces the ordinary white event.
PAL-002 — Seal Echo coexistence and consumption High
Requires: Level-40+ Twist of Light, both relevant seals and aura/event logging.
Observe / verify: Replace seals before the next auto, record simultaneous Echo auras, then compare landed and avoided autos plus a Judgement-only control.
PAL-003 — Windfury Totem proc and charge rules High
Requires: Level-32+ Shaman providing Totem, Paladin recipient, no weapon imbue or other extra-attack source.
Observe / verify: Count eligible autos, specials and seal events; record AP aura charges and the next normal swing. Isolate avoided events and exclude immediate recursion before estimating chance.
PRI-001 — Shadow Word: Death script value 150 High
Requires: Death rank one at level 32; fixed SP, talents and target mitigation.
Observe / verify: Compare damage and backlash above/below execute thresholds, lethal/nonlethal hits and cooldown behavior. Subtract ordinary Early Demise and damage-range effects before attributing anything to 150.
WAR-001 — Level-60 outgoing rage normalization High
Requires: Level 60 and known one- and two-handed base speeds.
Observe / verify: Measure uncapped ordinary swings while varying damage independently; test the existing 4.5 and 4.5/1.3 rates rather than fitting a damage formula anew.
WAR-002 — Off-hand and extra-swing rage High
Requires: Relevant talent ranks and the actual raid Windfury buff.
Observe / verify: Repeat the off-hand ratio with the rage talent and test proc-generated swings, independently of damage and crit.
WAR-006 — Flurry amount and charge events High
Requires: Warrior with 2/5-5/5 Flurry (27-30 talent points, beyond the level-30 beta).
Observe / verify: Swing intervals with each rank to confirm 5% attack speed per rank.
WAR-008 — Shield attacks and weapon procs High
Requires: Level-40 Shield Slam and access to the exact proc source, including level-32+ Windfury Totem where relevant.
Observe / verify: Repeat isolated attacks, separating landed/avoided outcomes and weapon versus shield effects.
WL-001 — Demonic Brand power source and threat High
Requires: Max-level pet ranks and relevant Tier effects.
Observe / verify: Verify the level extrapolation and Tier interactions rather than rediscovering the basic power owner.
WL-009 — Demonic Knowledge ranked values High
Requires: Higher-level Warlock.
Observe / verify: Confirm how the bonus scales with player level.
DATA-006 — Legacy item and enchant proc assumptions Medium
Requires: Actual Crusader/Fiery enchants, fixed-speed weapons and separate white-only, special-attack and extra-swing samples. Move earlier only if the beta supplies these high-skill enchants.
Observe / verify: Estimate probability by eligible event type and weapon speed; record haste, handedness and form. A triggered spell's coefficient does not establish its PPM.
DRU-010 — Feral stealth openers Medium
Requires: Level-32 Ravage and level-36 Pounce.
Observe / verify: Confirm opener damage and legal states against the implemented ranks.
HUN-011 — Hunter’s Mark rank-value anomaly Medium
Requires: Access to ranks three and four with fixed ranged AP and no competing Mark.
Observe / verify: Compare target-conditioned ranged AP/damage and tooltip values per rank; record the client build.
ROG-001 — Mutilate off-hand flat damage and Cold Blood Medium
Requires: Higher Mutilate ranks.
Observe / verify: Verify that the higher flat component follows the low-rank ordering; do not repeat already settled proc timing unless a patch changes it.
WAR-010 — Weaponmaster hand-specific armor ignore Medium
Requires: Level-60 armor-ignore combinations (items and debuffs not available in the beta).
Observe / verify: Confirm stacking with other armor-ignore effects.
WL-002 — Incinerate's Immolate bonus scope Medium
Requires: Level-40 Incinerate and two known SP totals on the same target.
Observe / verify: Measure with/without Immolate at each SP total. Compare the slopes, not just one tooltip-range average; retain identical damage talents and mitigation.
ROG-004 — Deadly Poison V acquisition Low
Requires: Level-60 Rogue and relevant trainer/vendor/content access if no source export resolves it first.
Observe / verify: Confirm acquisition without an unavailable raid book and record the obtainable rank.
SHA-011 — Totem of the Storm Maelstrom Weapon chance Low
Requires: Level-60 Enhancement Shaman with Maelstrom Weapon and Totem of the Storm.
Observe / verify: Maelstrom Weapon stacks gained per Lightning Bolt over about 100 casts.

Sources / code · 11 questions

Client data, source research, code inspection or simulator testing; no game experiment is required.

CORE-011 — Base attributes and class conversions High
Requires: An authoritative matching-build base-stat/conversion table, if obtainable.
Observe / verify: Compare exact level-60 rows and units to engine values; CombatRatings alone does not provide them.
DATA-002 — Projected gear budget and stat prices High
Requires: A proposed alternative budget or additional sourced crafted/PvP items.
Observe / verify: Fit out-of-sample items and compare total selected stat budgets by role; do not judge equivalence from the final DPS ranking.
PAL-008 — Vindication's target AP reduction High
Requires: Trait curve 82984 and child 440667 level scaling plus existing AP-debuff exclusivity.
Observe / verify: Reconcile the rendered level-60 amount with raw level fields; test activation/expiry and competing debuffs against known target AP.
SCEN-001 — Tank boss damage and external healing High
Requires: A chosen workload or credible raid-log encounter specification.
Observe / verify: Change damage/healing assumptions as a separately labeled scenario and compare matched inputs, rather than silently redefining the tank benchmark.
SCEN-003 — Tank target selection and threat coverage High
Requires: A three-target fixture with deliberately unequal per-target threat and debuff durations.
Observe / verify: Check selection, switching, maintenance and minimum target TPS; ensure current-target lookups do not accidentally read the first target's aura.
CORE-009 — Eureka cast and periodic bonuses Medium
Requires: Current engine, a Gnome DoT caster and a channel caster.
Observe / verify: Record tick damage with Eureka before application, after application, and on the last charge. Establish which effects the current wrapper actually preserves.
DATA-006 — Legacy item and enchant proc assumptions Medium
Requires: Crusader enchant 1900 → spell 20007; fixed proc events at levels 20 and 60.
Observe / verify: Check 100 Strength, the sourced self-heal range, health capping and damage/threat separation. Verify Lifestealing's heal separately if that enchant is enabled.
DRU-010 — Feral stealth openers Medium
Requires: Current learned ranks/effects and an explicit precombat stealth state.
Observe / verify: Implement legal form/position/stealth gates, combo points and one-use opener behavior; ensure normal combat cannot repeat stealth attacks.
SCEN-005 — General absorb-pool and shielding threat Medium
Requires: A shield with a sourced amount/duration and sequential controlled incoming hits.
Observe / verify: Verify partial and full consumption, expiry, overwrites and ordering without converting unconsumed capacity into effective healing.
SHA-004 — Owned buff-totem effects and replacement Medium
Requires: Owned low/high-rank totems plus independent external providers.
Observe / verify: Check actual stats/resource returns, replacement, expiry and recall. Removing an owned provider must not delete a surviving external provider or leave its own effect permanent.
ROG-004 — Deadly Poison V acquisition Low
Requires: Current trainer/vendor/recipe or loot-source export.
Observe / verify: Join the learned recipe to an obtainable source; distinguish the crafted poison item from its recipe.

Source-supported implementation follow-up · not applied

These specific changes do not need an unknown server rule to be invented. Some records also contain separate unresolved behavior.

DATA-010 — Raw haste rating conversion: Represent raw haste rating separately from flat percentages, using the sourced level-60 conversion of 10 rating per percent.
DATA-011 — Item-picker shared-crit EP: Project shared equipment crit/hit into picker valuation exactly once, preserving explicitly scoped exceptions.
DATA-008 — Unimplemented vendor item effects: Implement Defender's Grip's ordinary outside-city 8% Block for 15 seconds; do not invent the other items' script behavior.

Shared combat and resources 6 active / 22 total

CORE-001 Boss attack tables Needs evidence · high

Question: What are level-60 miss, dodge, parry, crit-suppression and glancing rules against level-63 targets?

Partly answered · moderate confidence: An October 3 public measurement has 300 misses/4259 attacks against+3 targets at 1% hit:7.04%,95% Wilson about 6.31–7.85%. It supports an 8% rather than 9% single-weapon/special cap. This supersedes treating all suppression behavior as Classic-confirmed; ranged/dual-wield/boss-level scopes are not established by this sample. October 11 audit: Two sources agree that hit suppression is gone: the October 3 measurement (300 misses in 4,259 attacks at 1% hit against +3 targets, 7.04%, 95% CI 6.31–7.85%, excluding the 8% a suppressed table predicts) and magey/forever-warrior#1's completed checklist item 'Hit suppression is gone'. The same issue's October 10 sample (3,644 attacks) confirms crit-aura suppression (5.19% observed against 5.61% expected) and Classic miss 8%, dodge 6.5%, parry 14%, glancing 40% and block 5% rates against +3 targets. It also reports glancing damage reduction as currently broken in the beta (about 24.6% observed), so that number is not adopted.

Current implementation / scenario: Forever removes Classic hit suppression for melee and ranged weapon attacks: the 8% special miss against +3 targets falls with every point of hit. Crit-aura suppression (1.8% against +3) and Classic dodge, parry, block and glancing chances remain; glancing damage keeps the Classic formula while the beta's glancing penalty is acknowledged as broken.

Impact: All melee, ranged and tank comparisons; hit caps and crit value.

Remaining scope: Glancing damage once Blizzard fixes it; two-roll special crit; bonus weapon skill; level-63 confirmation.

Implementation / model follow-up: Implemented: sim/core/target.go applies HitSuppression only outside Forever. Profiles selected for the old 9% cap are now 1% over the true 8% cap; gear re-selection is required to realise the gain.

Answer routes

Current beta · level 30
Requires: Known skill/hit, an equal-level and a +3-level target, recorded facing.
Observe / verify: Count white/special/ranged outcomes separately; compare queued and unqueued attacks. This supplies a low-level table, not a level-63 certification.
Beyond level 30 / launch
Requires: Level-60 characters and level-63 targets on a build without the known glancing defect.
Observe / verify: Repeat outcome and crit-suppression counts; compare damage of glances with ordinary hits at fixed AP.

sim/core/attack.go · sim/core/spell_result.go · docs/check_dispositions.md · https://github.com/magey/forever-warrior/issues/1 · docs/october-8-review.md

Earlier references: T01 / T06

Link to CORE-001

CORE-002 Energy regeneration and haste Needs evidence · high

Question: Does general haste increase Energy regeneration, and do attack-speed-only effects leave it unchanged?

Game test needed · low confidence: PowerType establishes the base 10 Energy/sec, not the haste coefficient. The previously analyzed 1% sample spans both 10 and 10.1/sec within endpoint quantization error. No new controlled source in the reviewed issue collections settles the distinction.

Current implementation / scenario: 10 Energy/sec integrated in 100-ms steps; general haste provisionally increases the rate. Slice and Dice, Berserking and attack-only enchants do not. The step size is not a measured server tick.

Impact: Feral and all Rogue builds; haste and resource value.

Remaining scope: General-haste scaling, distinct from attack-speed effects.

Implementation / model follow-up: Keep the existing haste multiplier explicitly provisional; do not import retail resource rules as a fact.

Answer routes

Current beta · level 30
Requires: Two known general-haste totals, identical resource talents, uncapped recovery windows and precise resource timestamps.
Observe / verify: Repeat matched recovery cycles from low Energy to below the cap, with no attacks or refunds. Aggregate endpoint uncertainty; a single rounded 1% window is insufficient. Run an attack-speed-only control separately.

docs/energy_audit.md · docs/check_dispositions.md · sim/core/energy.go · https://wago.tools/db2/PowerType?build=1.60.1.69977

Earlier references: T15 (haste/cadence) / T35

Link to CORE-002

CORE-004 Spell GCD versus physical GCD Model assumption · high

Question: Which speed effects shorten magical and physical GCDs, and what is the minimum?

Credible working answer · moderate confidence: The consistently applied physical-school versus magical-school rule is a reasonable accepted working model, corroborated by the pinned WoWSims parser. The client gives category 133 and 1500 ms for both shocks and Mortal Strike, so those fields cannot independently prove which GCDs scale. This is not a remaining per-spell implementation hunt.

Current implementation / scenario: The adopted WoWSims-style rule hastes magical GCDs with casting speed, retains fixed physical GCDs, and uses a one-second floor. Missing magical schools were corrected across the audited utility spells.

Impact: All rotations; especially instant Shaman spells and physical 1.5-second abilities.

Remaining scope: An optional empirical check of the shared rule and one-second floor; not separate tests for every spell with the same classification.

Implementation / model follow-up: Keep the adopted rule and its regression coverage. Preserve the distinction between damage school and melee/ranged hit table.

Answer routes

Current beta · level 30
Requires: Known casting-speed modifier, one magical instant and one physical 1.5-second action; client cooldown timestamps.
Observe / verify: Compare GCD duration independently of cast time. Only test the floor when enough haste is actually available; a 1% modifier cannot establish it.

docs/speed_audit.md · docs/weekly_review.md · sim/core/cast.go · sim/core/spell.go

Earlier references: T17 / T40 (GCD)

Link to CORE-004

CORE-007 Judgement of Wisdom trigger rules Needs evidence · high

Question: Which events trigger Wisdom, at what probability and with what cooldown?

Game test needed · low confidence: The mana amount is sourced: effect 698050 returns 59 at the highest rank. The proc probability is not settled by Wowhead's displayed 50%: all three captured SpellAuraOptions versions instead have 100% on the parent, which triggers dummy intermediate spell 1826. A server filter can exist behind that intermediate. Neither field proves the complete proc rule.

Current implementation / scenario: The level-60 return is 59 mana. The inherited event/proc model, including its 50% chance, remains provisional; self-hits are excluded.

Impact: Mana sustainability of melee Hunters, Paladins and other frequent attackers.

Remaining scope: Effective proc chance, avoided-white eligibility, periodic/proc eligibility and ICD.

Implementation / model follow-up: Keep 50% explicitly provisional; do not replace it with 100% merely from the parent. The current code even allows avoided white swings, which needs its own event-category check.

Answer routes

Beyond level 30 / launch
Requires: A level-38+ Paladin with Seal/Judgement of Wisdom, uncapped mana recipients and isolated event sources.
Observe / verify: Count mana returns per white hit, avoided white, special, direct spell, periodic tick and triggered effect. Compare event intervals to distinguish chance from ICD; the level-30 cap does not grant this seal.

docs/check_dispositions.md · sim/core/debuffs.go · https://www.wowhead.com/forever/spell=20186/judgement-of-wisdom · https://wago.tools/db2/SpellAuraOptions?build=1.60.1.69977 · https://wago.tools/db2/SpellEffect?build=1.60.1.70009

Earlier references: T55

Link to CORE-007

CORE-011 Base attributes and class conversions Needs evidence · high

Question: Are level-60 naked attributes and attribute-to-crit/resource conversions correct for every supported race/class?

Partly answered · moderate confidence: Rating conversions are not all unknown: the captured level-60 CombatRatings table supplies 14 crit rating and 10 hit/haste rating per percentage point. Those are distinct from Agility/Intellect conversion slopes and naked race/class base attributes. The latter are still inherited and are not established by the rating file or by low-level character examples. October 11 audit: PlayerExpectedStat (70338) confirms the level-60 conversions for all nine classes exactly: crit per Agility (Warrior 0.0500%, Paladin 0.0506%, Hunter 0.0189%, Rogue 0.0345%, Priest 0.0500%, Shaman 0.0508%, Mage 0.0514%, Warlock 0.0500%, Druid 0.0500%), spell crit per Intellect, base mana (e.g. Paladin 1512, Hunter 1720, Mage 1213) and 10 health per Stamina all match base_stats.go.

Current implementation / scenario: Retained class/race tables with explicit Forever racial changes. Another fork's level-70 attributes were not imported.

Impact: Every comparison; physical Agility crit versus caster Intellect crit and tank health.

Remaining scope: Naked level-60 base attributes per race/class; PlayerExpectedStat does not carry them.

Implementation / model follow-up: Do not use equal item budgets or a correct generic-crit import as proof that base stat tables are correct. Avoid importing another fork's level-70 rows.

Answer routes

Sources / code
Requires: An authoritative matching-build base-stat/conversion table, if obtainable.
Observe / verify: Compare exact level-60 rows and units to engine values; CombatRatings alone does not provide them.
Beyond level 30 / launch
Requires: Naked level-60 characters with racial/talent bonuses accounted for, plus isolated Agility/Intellect gear deltas.
Observe / verify: Record base attributes, health/mana and crit slopes for each class; race pairs isolate racial offsets.

sim/core/base_stats.go · docs/upstream-forever-review-2026-09-23.md · docs/crit_model.md · assets/db_inputs/forever_combat_ratings.json

Earlier references: T51 in in_game_checks.md (base attributes)

Link to CORE-011

CORE-009 Eureka cast and periodic bonuses Needs evidence · medium

Question: Does the last Eureka charge benefit every tick of the consuming channel/DoT after the aura expires?

Partly answered · moderate confidence: October 1 excludes periodic damage/healing from Eureka. The ordinary-DoT snapshot/application-tick leak is reproduced and repaired, and same-spell channel coverage is checked. Historical reports of dynamic DoT bonuses remain preserved as pre-fix behavior, not intended rules. October 15 update: The October 15 notes and client (1259812-1259823, duration 15 s to 8 s) remove the charges, so the last-charge question no longer applies.

Current implementation / scenario: Eureka is an 8-second buff with no charges (client 1.60.1.70425): every eligible direct cast inside the window gets the cost and damage bonus. Ordinary periodic damage stays excluded; a channel started inside the window keeps the bonus on its ticks.

Impact: Gnome resource use and damage, particularly channels.

Remaining scope: Whether a channel started just before expiry keeps the bonus on later ticks.

Implementation / model follow-up: Focused periodic-application, same-spell channel and nested-charge regressions pass. Do not relabel the old snapshot assumptions as current game verification.

Answer routes

Sources / code
Requires: Current engine, a Gnome DoT caster and a channel caster.
Observe / verify: Record tick damage with Eureka before application, after application, and on the last charge. Establish which effects the current wrapper actually preserves.
Current beta · level 30
Requires: Gnome Mage/Warlock/Priest with a low-rank DoT or channel; exact build and buff-stack timestamps.
Observe / verify: Repeat those three cases in game. Do not infer intended behavior from an open bug report or conflate the old mana-discount amount with current values.

docs/forever_races.md · sim/core/forever_gnome.go · https://github.com/ClassicWoWCommunity/forever-bugs/issues/35 · https://github.com/ClassicWoWCommunity/forever-bugs/issues/90 · https://wago.tools/db2/SpellEffect?build=1.60.1.70009 · sim/core/spell_result.go · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-1/2360696/4 · docs/october-1-corrections.md · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6

Earlier references: T03

Link to CORE-009

CORE-003 Energy refunds on avoided attacks Accepted working answer · high

Question: Which builders and finishers refund Energy on miss, dodge or parry, and what fraction of the paid cost returns?

Credible working answer · moderate confidence: The earlier review misread the client. Every player rank of Eviscerate (2098–31016), Rupture, Slice and Dice, Expose Armor, Rip (1079–9896) and Ferocious Bite (22568–31018) carries SPELL_ATTR1_DISCOUNT_POWER_ON_MISS (0x08000000) in 70009 and 70338. The ids without it are cost-free NPC or internal records with no skill line (1271520, 1271514, 1272982, 1289141, 1289142). In Era only builders carry the bit, and builders refund 80%; Forever extends the same attribute to finishers. Primal Bite, Claw, Rake, Shred, Maul and Lacerate have it; Swipe and the retained Mangle ids do not. October 11 audit: Rogue beta testers corroborate the finisher refund: 'parries or dodges refund cps and most energy' and 'energy refunds are in for everything' (Rogue Classic Discord, 2026-09-28/29). The same tester said misses do not refund, which differs from Classic's builder rule and from the engine (which refunds on any avoided outcome); that single statement is recorded, not adopted.

Current implementation / scenario: Forever finishers refund 80% of their paid Energy on a miss, dodge or parry, like builders: Eviscerate, Rupture, Expose Armor, Rip and Ferocious Bite. Combo points are spent only on a landed finisher.

Impact: Feral and Rogue resource supply, especially before hit cap or when attacking from the front.

Remaining scope: Whether a miss (as opposed to dodge/parry) refunds, and the exact fraction ('most').

Implementation / model follow-up: Implemented: Forever finishers use Refund 0.8 (eviscerate.go, rupture.go, expose_armor.go, rip.go, ferocious_bite.go).

Optional / conditional follow-up

Current beta · level 30
Requires: A Rogue with builders/Eviscerate/Rupture, or a level-20 Cat with Claw/Rake/Rip; uncapped Energy and an avoidance-producing target.
Observe / verify: Pair resource changes and combo points with each avoided action; subtract independently measured regeneration. Test builders and finishers separately.
Beyond level 30 / launch
Requires: Level-32+ Cat with Ferocious Bite.
Observe / verify: Separate initial cost, excess-Energy spending, refund and combo-point retention on avoided Bites.

docs/energy_audit.md · sim/core/energy.go · sim/rogue/eviscerate.go · sim/druid/ferocious_bite.go · https://wago.tools/db2/SpellMisc?build=1.60.1.70009

Earlier references: T16 / H09

Link to CORE-003

CORE-005 Channel speed sources Resolved · high

Question: Which general-haste effects compress channels, distinct from casting-speed-only effects?

Source-backed answer · moderate confidence: No Forever channel carries SPELL_ATTR5_SPELL_HASTE_AFFECTS_PERIODIC (0x2000): Drain Life, Drain Soul, Arcane Missiles, Mind Flay, Starshards and all four Penance ranks lack it, and Penance's Era id had it before Forever removed it. In the retail-derived server (TrinityCore Unit::ModSpellDurationTime), channel duration is hasted only with that attribute or the melee-haste equivalent. This agrees with the Berserking/Arcane Missiles observation.

Current implementation / scenario: Channels are not hasted in Forever. The separate SpellHaste channel input has no current source and stays dormant.

Impact: Mage, Priest and Warlock channels; hypothetical haste scaling.

Remaining scope: None for current gear; reopen only if a channel gains the attribute.

Implementation / model follow-up: No code change needed while no SpellHaste source exists; a future raw haste source must not shorten channels.

Optional / conditional follow-up

Current beta · level 30
Requires: Arcane Missiles and separately isolated Berserking or Wind Blessed; all missile timestamps.
Observe / verify: Compare complete channel duration and tick spacing with a no-modifier control, excluding Missile Barrage.
Sources / code
Requires: A concrete obtainable general-haste effect or rating-bearing item.
Observe / verify: Identify its aura types before scheduling a gameplay test. Acquisition is not established merely by the hypothetical SpellHaste stat; move to launch only if the chosen source requires it.

docs/speed_audit.md · docs/check_dispositions.md · sim/mage/arcane_missiles.go · sim/priest/mind_flay.go

Earlier references: T17 / T40 (channels)

Link to CORE-005

CORE-008 Dynamic periodic damage boundaries Resolved · high

Question: Which damage components update during an existing DoT, and which cast-specific bonuses remain attached?

Partly answered · moderate confidence: The Forever Hunter wiki explicitly describes dynamic Sting RAP and periodic crit. Public reports also distinguish intended dynamic Thorns/Ret Aura scaling from their beta snapshot bug. These support the ordinary no-snapshot rule, but not every charged-cast or proc-payload exception.

Current implementation / scenario: Forever periodic damage uses dynamic stats/modifiers under the adopted no-snapshot model, with explicit cast-specific effects and Ignite exclusions. Deep Wounds also reads current weapon/AP state.

Impact: All DoT builds; temporary buffs, weapon swaps and poison upkeep.

Remaining scope: Narrow cast-specific exceptions belong to Eureka, Inner Focus and the individual proc records, not a blanket retest of every ordinary DoT.

Implementation / model follow-up: Keep ordinary dynamic periodic damage as the reference rule. Do not multiply an already calculated Ignite/Deep Wounds payload again merely because ordinary DoTs are dynamic.

docs/check_dispositions.md · sim/core/dot.go · sim/warrior/deep_wounds.go · https://github.com/classic-hunter/forever-hunter/wiki/Forever-Beta-Changes · https://github.com/ClassicWoWCommunity/forever-bugs/issues/117

Earlier references: T52 in in_game_checks.md (dynamic DoTs) / T02 Deep Wounds follow-up / T35 periodic scope

Related: CORE-009 · PRI-003 · PAL-010

Link to CORE-008

CORE-014 Mana restoration threat Outside current scenarios · high

Question: Which mana returns generate threat, who receives it, and is it divided between enemies?

Outside current scenarios · moderate confidence: Threat from mana gains is a negligible part of tank threat in the current scenarios. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Explicit NoThreat covers Forever Life Tap owner/demon, Dark Sacrifice, Shield Specialization and Improved Seal of Fury mana returns. Amounts/proc gates remain separately sourced; ordinary gains retain existing accounting.

Impact: Tank TPS and rotation comparisons: wasting mana can increase subsequent effective restoration and its credited threat without improving damage or survival.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Current beta · level 30
Requires: A mana user, effective mana potions, threat API readings and one versus several engaged enemies.
Observe / verify: Measure total and per-enemy threat from one isolated return, including same-type and mixed-type packs. A mana-capped use is the wasted-resource control.

sim/core/mana.go · sim/core/spell.go · https://github.com/ClassicWoWCommunity/forever-bugs/issues/66 · docs/october-8-review.md · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/paladin/sof.go · sim/paladin/talents.go · tools/forever_bench/paladin_october9_test.go

Related: CORE-007 · SCEN-003 · PAL-006

Link to CORE-014

CORE-016 Common player Spirit regeneration Accepted working answer · high

Question: Do all player classes share the Mage/Priest formula, including its intercept?

Credible working answer · moderate confidence: Both Hunter pairs and five of six Paladin pairs fit 6.25 + Spirit/8, apart from roughly 0.001 in API output. The first Paladin pair would fit at 48 rather than its reported 49 Spirit; this single outlier does not establish a different Paladin intercept. The API dump omitted effective-Spirit and casting-regeneration values. A common-player formula is a better-supported provisional model, not proof for all classes or level 60.

Current implementation / scenario: Forever players use the common working formula 6.25 + effective Spirit/8 MPS. Explicit class overrides, pets, Classic regeneration and the separate MP5 contribution remain unchanged.

Impact: Mana pacing, talents and gear for non-Mage/Priest mana users.

Remaining scope: Effective versus first-return UnitStat values, fractional stats, levels, MP 5/buff controls and coverage of other classes/pets.

Implementation / model follow-up: Implemented the provisional common-player formula without an API-offset fudge. Focused checks cover the consistent measurements, casting/noncasting Spirit, both MP5 modes, preserved explicit overrides and unchanged pet/Classic formulas. Refresh affected mana pacing and comparisons; retain the unmatched raw sample.

Optional / conditional follow-up

Current beta · level 30
Requires: Hunter/Paladin plus another mana class, several Spirit values, no MP5/buffs, sufficient mana deficit.
Observe / verify: Capture all UnitStat returns, GetManaRegen values, build/level/race and timed actual regeneration both inside/outside five-second rule.
Beyond level 30 / launch
Requires: Unresolved higher-level classes/passives.
Observe / verify: Repeat only remaining level/class terms rather than assume low-level intercept is universal.

docs/october-8-review.md · sim/core/mana.go

Link to CORE-016

CORE-019 Talent modifier scope versus client class masks Resolved · high

Question: Do simulated talents apply their damage, crit, hit and cost modifiers to the same spells as the client's class masks?

Source-backed answer · high confidence: Differential audit, October 11: each benchmark build was constructed with and without every talent it takes (583 variants). The spells whose static modifiers changed were compared with the spells covered by the talent's client modifier masks (aura 107/108, ops damage, periodic, crit, crit damage, hit, cost, cast, cooldown and GCD). Defects found and fixed: Twin Disciplines applied to every instant Priest spell, including Mind Flay, Shadow Word: Death, Starshards and Penance, while the mask covers Shadow Word: Pain, Devouring Plague and Holy Nova; a tester also found it does not affect Mind Flay. Holy Precision, Holy Specialization and Mental Agility applied to Penance (and Mental Agility to Shadow Word: Death), which their masks exclude. Mortal Shots applied to Sniper Shot, Raptor Strike, Mongoose Bite and Wing Clip but not Serpent Sting; the mask is Auto/Aimed/Multi/Arcane/Serpent/Volley. Impale did not reach Rend's crits, which its mask includes. Investigated and correct: Elemental Fury on Searing/Magma totems (the mask includes their attack spells), Aftermath and Improved Seals (applied at cast or on the proc spell). Low-impact mismatches recorded but not changed: Genesis and Survival Tactics on AoE-only spells, Call of Flame on Fire Nova, Malediction's Soul Fire scope, and Paladin seal cost reductions (Paladins are never mana-limited in the benchmark).

Current implementation / scenario: Talent modifiers follow the client masks for every benchmark talent checked; exceptions are listed in the answer.

Impact: Shadow and Smite Priest, Marksmanship and Beast Mastery, and Warriors using Rend.

Remaining scope: Dynamic (aura-based) talent effects are outside a static differential check.

Implementation / model follow-up: Fixed under Forever only; Classic behaviour unchanged.

sim/priest/talents.go · sim/hunter/talents.go · sim/hunter/serpent_sting.go · sim/warrior/rend.go

Link to CORE-019

CORE-020 Critical damage racials (Tauren Endurance, Dwarf Earth Fury) Accepted working answer · high

Question: Does the 2% critical damage bonus scale the whole critical multiplier or only the bonus part?

Credible working answer · moderate confidence: Client 1.60.1.70425: Endurance 20550 e1 aura 163 (2) and e2 aura 50 (2) replace the 1% hit; Earth Fury replaces Mace Specialization. Aura 163 scales the full critical damage in the server code this client descends from.

Current implementation / scenario: Scales the whole multiplier, like client aura 163: melee crits 2.04x, spell crits 1.53x.

Impact: Every Tauren and Dwarf profile, and the rotation planner for Tauren characters.

Remaining scope: A measured Tauren or Dwarf crit/non-crit ratio.

Implementation / model follow-up: sim/core/racials.go (critDamageAura); sim/planner/warrior_model.go reads it from the attack table.

Optional / conditional follow-up

Current beta · level 30
Requires: Tauren or Dwarf Warrior and a non-Tauren control with the same weapon.
Observe / verify: Ratio of crit to non-crit white damage over about 300 swings each.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · sim/core/racials.go

Link to CORE-020

CORE-023 Class weapon imbues and temporary weapon enchants Resolved · high

Question: Can a Forever weapon carry a Shaman imbue or Rogue poison together with a stone, weightstone or oil?

Source-backed answer · high confidence: Client 1.60.1.70425: every Shaman imbue and Rogue poison rank applies its enchantment with effect 360, while stones, weightstones and oils use effect 54 (the temporary-enchant slot). Rogue and Shaman testers report poisons, stones and Windfury together since September. The 70291 notes already made weapon enchants work in forms.

Current implementation / scenario: Yes. Shaman imbues and Rogue poisons are class options; stones, weightstones and oils stay in the Consumes weapon-enchant slot, on both hands. Classic keeps one shared temporary-enchant slot.

Impact: Enhancement (Windfury Weapon plus Shadow Oil), Rogues (poisons plus oils or stones), Feral and Bear (stones work in forms).

Remaining scope: None.

Implementation / model follow-up: Shaman spec options main_hand_imbue/off_hand_imbue (Forever); the web shows the class controls under Forever and stones and oils for Feral.

sim/shaman/shaman.go · sim/rogue/rogue.go · sim/core/consumes.go

Link to CORE-023

CORE-006 Optional MP5-per-second interpretation Accepted working answer · medium

Question: Does a flat displayed MP5 point restore one mana per five seconds or per second?

Credible working answer · high confidence: The beta report supplies 0/2/4 gear-MP5 controls, a combat-log attachment and video; a project contributor acknowledges an engineering bug. It explicitly reports normal five-second behavior for Blessing of Wisdom and Mageblood. This is evidence for a gear-specific beta defect, not a universal redefinition of MP5.

Current implementation / scenario: Ordinary MP5/5 per second is the default. The explicit optional mode multiplies only the MP5 stat contribution by five; it is off in rankings.

Impact: Mana users if enabled; does not establish a current ranking error.

Remaining scope: Whether the gear bug has been deployed as fixed on a later build. That is a beta-version watch, not a prerequisite for the intended ordinary-MP5 benchmark.

Implementation / model follow-up: The default MP5/5 rule remains appropriate for the intended model. The optional uniform fivefold setting is a hypothetical sensitivity mode, not a faithful emulation of the reported gear-only bug; its description should say so.

Optional / conditional follow-up

Current beta · level 30
Requires: Only if reproducing a particular live beta build is needed; gear with a known MP5 delta and a separate Wisdom/Mageblood control.
Observe / verify: Compare uncapped regeneration while continuously spending mana. Keep Spirit fixed and record the exact client build; distinguish item MP5 from buff MP5.

docs/mana_regeneration.md · sim/core/mana.go · https://github.com/ClassicWoWCommunity/forever-bugs/issues/54

Earlier references: T44

Link to CORE-006

CORE-010 Read Ley Line regeneration Outside current scenarios · medium

Question: Which regeneration components does High Order's active ability increase, and does location matter?

Outside current scenarios · moderate confidence: No benchmark rotation uses Read Ley Line. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: The active regeneration effect is not used; a nearby ley line is not assumed.

Impact: Potential High Order mana recovery; currently absent rather than a free benefit.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Current beta · level 30
Requires: High Order mana user; fixed Spirit, a known MP5 item delta, uncapped mana and no other temporary regen.
Observe / verify: Compare idle and continuous-spending recovery with the short buff. Record application/expiration and test the long location buff separately if needed.

docs/forever_races.md · sim/core/racials.go · https://www.wowhead.com/forever/spell=1259705/read-ley-line · https://us.forums.blizzard.com/en/wow/t/alliance-skybourne-racial-feedback/2362168 · https://wago.tools/db2/SpellEffect?build=1.60.1.70009

Earlier references: T04

Link to CORE-010

CORE-012 Tea and conjured-item shared cooldowns Outside current scenarios · medium

Question: Which consumables share cooldowns with Thistle Tea, including prepull uses?

Outside current scenarios · high confidence: All 31 current Tea profiles select Tea as their sole conjured item; the registration switch allows one selected conjured consumable. Their five-minute scenario has no Tea prepull, and Tea's own five-minute cooldown prevents another use within the fight. The speculative cross-conjured cooldown does not constrain these requests.

Current implementation / scenario: Rogues and Cats can use the sourced 100 Energy, five-minute Tea. The inherited cross-conjured shared-category handling is retained.

Impact: Feral/Rogue resource plans and mixed consumable configurations.

Remaining scope: Shared-category behavior matters only for a future mixed-conjured or different prepull setup.

Implementation / model follow-up: Keep the sourced Tea amount and personal cooldown. Move the cross-item interaction out of the current benchmark test queue; do not call its actual server category measured.

docs/mechanics_review.md · sim/core/consumes.go · artifacts/forever_input_profiles.json

Earlier references: T24 (shared cooldown portion)

Link to CORE-012

CORE-015 Weapon-racial crit and ranged attacks Outside current scenarios · medium

Question: Do weapon-specialization racials affect ranged crit, or only melee, spell and inherited pet crit?

Outside current scenarios · moderate confidence: Client 1.60.1.70425 replaces the weapon racials: Human Ruthless (+2% crit with all spells and attacks), Orc Savage (+1%), Dwarf Earth Fury (+2% critical damage). None depends on the held weapon, so the ranged-scope question no longer arises.

Current implementation / scenario: A matching held weapon adds MeleeCrit and SpellCrit; the ranged attack path also reads MeleeCrit.

Impact: Could change Hunter racial comparisons and gear choices without changing generic equipment crit.

Remaining scope: None.

Implementation / model follow-up: sim/core/racials.go.

Optional / conditional follow-up

Current beta · level 30
Requires: Human or Orc Hunter and matching versus nonmatching melee weapons with equal Agility/crit stats; unchanged ranged weapon and pet.
Observe / verify: Record melee, ranged, spell and pet crit APIs before/after the swap, then compare ranged attack outcomes if the display and source descriptions disagree. Check one and two matching weapons separately rather than assuming the bonus stacks.

https://github.com/classic-hunter/forever-hunter/wiki/Forever-Beta-Changes · sim/core/specializations.go · sim/core/racials.go · sim/core/spell_result.go · sim/core/spell_outcome.go · artifacts/question_review/scenario_data_evidence.json · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6

Related: HUN-001 · DATA-012

Link to CORE-015

CORE-017 Crowd-control DR target policy and reset anchor Outside current scenarios · medium

Question: Which Forever targets receive Root group 1/proc Stun group 10 DR, and does its15-second reset begin at application or final effect end?

Outside current scenarios · high confidence: No benchmark scenario uses crowd control. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Target stun/root/slow eligibility and DR are explicit opt-ins. Source groups use15s/×.5/immunity after3; default after-final-end versus alternate after-application is exposed. WingClip group0/slows bypass DR.

Impact: Opt-in Hunter/Mage control scenarios, not default boss damage.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Current beta · level 30
Requires: Level 30 eligible cooperating targets, Entrapment/Impact/Improved Concussive, precise effect timestamps and two casters.
Observe / verify: Compare repeated applications, breaks/overlaps and timings straddling15s from application versus effect end; include immune/no-DR controls.

https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/core/crowd_control.go · sim/core/movement.go · tools/forever_bench/hunter_utility_october9_test.go · https://wago.tools/db2/SpellDiminish?build=1.60.1.70291

Related: SCEN-009 · DATA-001

Link to CORE-017

CORE-021 Touch of the Grave after 1.60.1.70425 Accepted working answer · medium

Question: What are the new proc chance and cooldown?

Credible working answer · moderate confidence: The October 15 notes give 30%/20% and a 25-second internal cooldown and remove Shadow scaling; the client aura options changed with them.

Current implementation / scenario: 20% (Warrior, Paladin, Rogue) or 30% (casters) on a landed damaging hit, 25-second cooldown, 5% of maximum health, no Shadow scaling.

Impact: Undead profiles.

Remaining scope: Whether the cooldown starts on the proc or on any eligible hit.

Implementation / model follow-up: sim/core/racials.go.

Optional / conditional follow-up

Current beta · level 30
Requires: Undead caster.
Observe / verify: Time between procs over 10 minutes of casting.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · https://wago.tools/builds-diff?to=1.60.1.70425&from=1.60.1.70374

Link to CORE-021

CORE-013 Physical multi-school mitigation Outside current scenarios · low

Question: When armor and magic resistance are similar, which mitigation does a physical multi-school spell select?

Outside current scenarios · high confidence: The current production spell registrations contain no Physical-plus-magic school combination; the explicit combination found is in a core test. Frostfire's two magic schools are not this case. The unresolved armor-versus-resistance threshold therefore has no demonstrated caller in the current benchmark.

Current implementation / scenario: The inherited threshold approximation acknowledges that choosing the numerically lowest value is insufficient.

Impact: Conditional multi-school physical effects; not a generic spell-resistance rewrite.

Remaining scope: Reopen when an actual supported Physical-plus-magic ability is introduced.

Implementation / model follow-up: Keep as a conditional engine capability issue, not a required gameplay experiment or a reason to alter single-school mitigation.

sim/core/spell_resistances.go · sim/core/spell_school.go · sim/core/spell_school_test.go

Link to CORE-013

CORE-018 Missing global cooldowns and Sweeping Strikes duration Resolved · low

Question: Do simulated spells use the client's GCD and aura durations?

Source-backed answer · high confidence: A spell-by-spell comparison of all 201 profiles' registered spells against 70338 (cast, cooldown, GCD, cost, durations, after client talent and Tier 1 modifiers) left these mismatches: Shield Wall 871, Intimidation 19577 and Stoneform 20594 have StartRecoveryTime 1500 in both Era and Forever but no GCD in the engine; Sweeping Strikes 12292 lasts 20 seconds (5 charges) in both clients, not 10.

Current implementation / scenario: Shield Wall and Intimidation trigger a 1.5-second GCD; Sweeping Strikes lasts 20 seconds. Stoneform's GCD is noted but it is never cast automatically.

Impact: Protection Warrior emergency timing, Beast Mastery GCD use, multi-target Arms.

Remaining scope: None.

Implementation / model follow-up: Implemented except Stoneform (manual-only cooldown).

sim/warrior/shield_wall.go · sim/hunter/talents.go · sim/warrior/sweeping_strikes.go

Link to CORE-018

Warrior 5 active / 22 total

WAR-001 Level-60 outgoing rage normalization Needs evidence · high

Question: Do the measured low-level speed-based rage rates persist at level 60?

Partly answered · high confidence: October 1 supersedes the earlier no-extra-critical-Rage beta answer: the final Warrior-specific note gives +100%, after an earlier +75% bullet. The tested low-level base-speed model remains separate from this source-supported critical multiplier.

Current implementation / scenario: Damage-independent base-speed Rage: 4.5 per weapon-second for two-handed, 4.5/1.3 for one-handed MH; outgoing white criticals now award twice normal Rage. Haste changes event frequency, not base speed.

Impact: All Warrior damage, gear selection and rotation decisions.

Remaining scope: Level-60 base rates and applicability to unusual extra swings remain unmeasured.

Implementation / model follow-up: Implement the final +100% critical multiplier; retain the qualified base-speed extrapolation and separate incoming-Rage model.

Answer routes

Level-30 beta
Requires: A known weapon retained across levels and no other rage gains.
Observe / verify: Optional extrapolation checkpoint: compare uncapped rage per landed swing before and after leveling. This does not certify level 60.
Beyond level 30 / launch
Requires: Level 60 and known one- and two-handed base speeds.
Observe / verify: Measure uncapped ordinary swings while varying damage independently; test the existing 4.5 and 4.5/1.3 rates rather than fitting a damage formula anew.

sim/core/rage.go · docs/check_dispositions.md · docs/flurry_review.md · https://github.com/tzcnt/forever-data/tree/main/raw-logs · https://ppach-warriorcompendium.share.connect.posit.cloud/rage.html · https://github.com/ClassicWoWCommunity/forever-bugs/issues/47 · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-1/2360696/4 · docs/october-1-corrections.md

Earlier references: T02 (outgoing rates) / M06

Link to WAR-001

WAR-002 Off-hand and extra-swing rage Needs evidence · high

Question: What rage do off-hand, avoided and proc-generated swings award?

Partly answered · high confidence: Ordinary low-level OH half-rate has log support. Current client-derived DWS ranks give 10% OH Rage per point, replacing the old 20%; Furious Precision separately gives 4/7/10% OH hit. Known-issue stale DWS hit is not treated as intended behavior.

Current implementation / scenario: Half-rate OH before current DWS +10% Rage per point; landed criticals multiply ordinary outgoing Rage by two. Avoided autos award none; extra-swing/max-level behavior remains qualified.

Impact: Fury, Windfury and weapon-proc value.

Remaining scope: Proc-generated extra-swing eligibility, avoidance exceptions and max-level extrapolation; measure DWS separately from critical swings.

Implementation / model follow-up: Keep ordinary OH evidence distinct from extra-attack assumptions. Apply DWS +10% per point and the separate OH-only hit talent.

Answer routes

Level-30 beta
Requires: A sourced extra-attack effect actually available by level 30, and separate MH/OH events.
Observe / verify: Isolate extra swings from regular autos and energizes. If no suitable effect is obtainable, defer that subtest rather than substituting ordinary swings.
Beyond level 30 / launch
Requires: Relevant talent ranks and the actual raid Windfury buff.
Observe / verify: Repeat the off-hand ratio with the rage talent and test proc-generated swings, independently of damage and crit.

sim/core/rage.go · docs/check_dispositions.md · https://ppach-warriorcompendium.share.connect.posit.cloud/rage.html · https://github.com/ppach/marrow.github.io/tree/master/eternal/data · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-1/2360696/4 · docs/october-1-corrections.md

Earlier references: T02 (off-hand/extra attacks/avoidance)

Related: WAR-001

Link to WAR-002

WAR-006 Flurry amount and charge events Needs evidence · high

Question: Which events proc, refresh and spend Flurry, and does the talent curve or triggered aura determine speed?

Game test needed · low confidence: The linked Flurry issue and current compendium chapter explicitly remain a model/testing plan. The active Warrior tree puts the first rank behind 25 points: level 35, with rank five at 39. Ordinary haste retaining rage per swing is supported elsewhere, but does not establish every Flurry charge rule. October 11 log audit: The public foreverlogs.gg reports contain no Warrior Flurry windows. They do show that Shaman Flurry, which has the same three client charges, ends on the first white swing after the crit that applied or refreshed it (SHA-009). Warrior Flurry may differ (the Shaman version also has a 500 ms consumption lockout), so the Warrior model keeps three charges until a Warrior is measured. Talent access: The level-30 beta gives 26 talent points, so a Warrior can take 25 Fury points (including Death Wish) and 1/5 Flurry now. The question matters: if Warrior Flurry lasts one white swing like Shaman Flurry (SHA-009), Fury and Fury (Sunder) lose about 7% and 2H Bloodthirst about 6% in the benchmark.

Current implementation / scenario: 5–25% melee speed; three shared charges. White-tagged events, including avoided swings and queued replacements, consume charges without a 500-ms gate. Melee crits refill after consumption. Each Cleave target may consume a charge.

Impact: All Flurry builds, especially dual wield and cleave.

Remaining scope: White/queued/extra-swing consumption, multi-target Cleave, refresh order and mid-swing timing.

Implementation / model follow-up: Retain the current disclosed model; no source justifies changing charge logic in this research pass.

Answer routes

Current beta · level 30
Requires: Level-30 Warrior with 26 Fury points including Death Wish and 1/5 Flurry (for example Fury 25053005040010100, no Arms or Protection points); advanced combat logging.
Observe / verify: Auto-attack a mob for three minutes with one weapon, then dual wielding, then with Heroic Strike queued every swing. Count white swings between each Flurry gain and fade, and note whether the buff icon shows 3 charges. About 60 Flurry procs per setup.
Beyond level 30 / launch
Requires: Warrior with 2/5-5/5 Flurry (27-30 talent points, beyond the level-30 beta).
Observe / verify: Swing intervals with each rank to confirm 5% attack speed per rank.

docs/flurry_review.md · sim/warrior/talents.go · https://github.com/magey/forever-warrior/issues/12 · https://ppach-warriorcompendium.share.connect.posit.cloud/flurry.html · ui/core/talents/trees/warrior.json · https://foreverlogs.gg/

Earlier references: T54

Link to WAR-006

WAR-008 Shield attacks and weapon procs Needs evidence · high

Question: Can Shield Slam and Shield Bash trigger main-hand effects such as Windfury?

Game test needed · low confidence: A competing implementation changing shield proc masks is not evidence of main-hand effect eligibility. The early test and the actual tank damage ability have different access: Shield Bash can be checked now, Shield Slam requires 40, and Windfury Totem requires a level-32 Shaman. A level-20 shield test cannot use Totem as its proc source.

Current implementation / scenario: Shield Slam is a main-hand special here; Shield Bash is not registered. Another implementation moved shield attacks off-hand, but its simulations do not prove server eligibility.

Impact: Protection DPS, rage and threat. Shield Slam requires later access.

Remaining scope: Shield-attack eligibility for specific equipped-weapon effects and Totem.

Implementation / model follow-up: Do not change Shield Slam's mask without an event-level source. Keep missing Shield Bash registration separate.

Answer routes

Current beta · level 30
Requires: Shield Bash, shield and an actually obtainable known weapon proc; autos stopped.
Observe / verify: Count proc events following isolated Bashes. This is a partial test, not certification of Shield Slam or Windfury.
Beyond level 30 / launch
Requires: Level-40 Shield Slam and access to the exact proc source, including level-32+ Windfury Totem where relevant.
Observe / verify: Repeat isolated attacks, separating landed/avoided outcomes and weapon versus shield effects.

sim/warrior/shield_slam.go · sim/warrior/warrior.go · docs/upstream_elliot_review.md · ui/core/talents/trees/warrior.json

Earlier references: T59

Related: WAR-012

Link to WAR-008

WAR-010 Weaponmaster hand-specific armor ignore Needs evidence · medium

Question: Can an off-hand mace grant the character-wide armor discount, and how does it stack?

Game test needed · low confidence: The active Weaponmaster tree grants its first point at level 30; the full five-point effect requires 34. The talent text's weapon conditions do not settle whether an off-hand mace grants an owner-wide armor discount. That hand-specific question is genuinely testable in the level-30 beta at one point. October 15 update: The October 15 notes also stop Weaponmaster's sword extra attack from triggering on Shield Bash and Shield Slam; the sim now excludes Shield Slam (Shield Bash is not a simulated attack).

Current implementation / scenario: The retained equipment gate can grant it from an off-hand mace. The competing upstream main-hand-only rule is not verified.

Impact: Mixed-weapon Warrior alternatives, not evidence for a free generic armor reduction.

Remaining scope: Hand scope and stacking of the mace armor-ignore effect.

Implementation / model follow-up: Keep the current gate provisional; do not replace it solely with another fork's main-hand-only rule.

Answer routes

Current beta · level 30
Requires: Level-30 Warrior with 1/5 to 5/5 Weaponmaster (20 Arms points first; 26 points are available), comparable mace and non-mace weapons, matched AP and controlled target armor.
Observe / verify: Move the mace between hands and measure main-hand damage changes; separate the weapon's own damage change and sword extra attacks. Repeat at 1/5 and 5/5 to confirm rank scaling.
Beyond level 30 / launch
Requires: Level-60 armor-ignore combinations (items and debuffs not available in the beta).
Observe / verify: Confirm stacking with other armor-ignore effects.

sim/warrior/talents.go · docs/upstream-forever-review-2026-09-23.md · ui/core/talents/trees/warrior.json · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6

Link to WAR-010

WAR-003 Incoming rage and mitigation Accepted working answer · high

Question: Does unmitigated damage × 10 / maximum health explain incoming rage at all armor/health values?

Scenario decision · moderate confidence: October 8 establishes expected creature health, shield-independent damage and correct crit/crush treatment, but not the coefficient, level key or exact normalization. The approved configurable central model and bounds are implemented and scoped callback/import checks pass; approval is not a server-formula measurement. October 11 audit: Developers' notes quoted in Fight Club (2026-10-08): 'Previously, we were balancing Rage gained around Warriors having 50% Armor ... We have rebalanced Rage gains based on appropriate values per level in the 20% to 40% range, varying by level.' A follow-up (2026-10-10) explains that the intended formula uses a level-based curve and that player maximum health was used by mistake for a while. This supports a level-based reference-armour model rather than player health; the exact curve is still unpublished.

Current implementation / scenario: Unreleased explicit provisional shared Warrior/Bear incoming model on the DIRECT hit-callback pipeline: (Damage + AbsorbedDamage) × coefficient / player-level ExpectedStat CreatureHealth / (1-reference Armor). Central coefficient 10 and clamped expected-Armor ratio 20–40% are assumptions; no actual-Armor undo or player-MaxHP denominator. Optional overrides preserve explicit zero. Periodic-only damage, including logical channel children represented as periodic, does not generate this modeled incoming Rage.

Impact: Protection rotation, self-damage resource gains and tank stat trade-offs.

Remaining scope: Exact coefficient, expected-health level selector, reference curve/operation and higher-level scaling remain unpublished. Periodic-damage eligibility is also unknown: dealDamageInternal dispatches periodic delivery only to OnPeriodicDamageTaken, while RageBar listens only to OnSpellHitTaken. Generic taking-damage wording does not verify all DoTs/channels or complete PvP support. Current tank workloads use direct swings/pulses.

Implementation / model follow-up: Use the exposed model and record resolved values/presence with each applicable row. Current tank selection/sensitivity and matrix evidence are complete; six coefficient20 guard failures remain prominent. Do not restore the old player-health/pre-Armor fit as verified or calibrate unknown values for selection gains.

Optional / conditional follow-up

Current beta · level 30
Requires: Level 30 Warrior and Bear, fractional Rage trace, no outgoing autos/spending/passive gains; independent Armor/Stamina and known enemy levels.
Observe / verify: Compare unabsorbed, partially/fully shielded ordinary/crit/crush hits; vary attacker level to identify the health lookup. A one-level fit establishes only an effective coefficient.
Beyond level 30 / launch
Requires: Level 60 and known incoming attacks, only when higher-level access exists.
Observe / verify: Validate extrapolation separately; do not treat client level 60 baselines as a live measurement.
Current beta · level 30
Requires: Level-30 Warrior/Bear, DoT-only attacker then separated channeled bolts; no outgoing autos, Bloodrage, Berserker Rage or other on-hit/resource bonuses. Preserve uncapped, unmodified fractional power tracing.
Observe / verify: Log actual damage, absorbed amounts and event types. Accumulate enough ticks for fractional Rage to be observable; compare isolated DoT/channeled delivery with a direct-hit control. Do not infer periodic eligibility from a rounded zero or mixed-resource combat.

sim/core/rage.go · docs/upstream-forever-review-2026-09-23.md · https://github.com/magey/forever-warrior/issues/3 · https://github.com/ClassicWoWCommunity/forever-bugs/issues/72#issuecomment-5777200313 · https://github.com/ClassicWoWCommunity/forever-bugs/issues/72#issuecomment-5808179814 · https://github.com/ClassicWoWCommunity/forever-bugs/issues/78#issuecomment-5786631000 · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/core/forever_incoming_rage.go · sim/core/forever_expected_stats.go · sim/core/forever_incoming_rage_october9_test.go · https://wago.tools/db2/ExpectedStat?build=1.60.1.70291 · sim/core/spell_result.go

Earlier references: T02 (incoming rage)

Related: DRU-012

Link to WAR-003

WAR-004 Berserker Rage's extra incoming rage Outside current scenarios · high

Question: Is the bonus a flat amount, a multiplier, or a damage-dependent rule?

Outside current scenarios · low confidence: The DPS boss never attacks DPS players, and the tank rotations do not use Berserker Rage. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Inherited +2 rage on landed melee/ranged hits and a separate damage-based spell-hit bonus; a proposed 2× rule was not adopted without evidence.

Impact: Incoming-damage Warrior scenarios and stance/cooldown choices.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Beyond level 30 / launch
Requires: Level-32+ Warrior, several known incoming hit sizes, fixed armor/health and no outgoing attacks.
Observe / verify: Compare rage with and without Berserker Rage separately for melee and spells; distinguish a constant bonus from a multiplier.

sim/warrior/berserker_rage.go · docs/upstream-forever-review-2026-09-23.md · https://wago.tools/db2/SpellEffect?build=1.60.1.70009 · https://wago.tools/db2/SpellLevels?build=1.60.1.69977

Earlier references: T02 (Berserker Rage)

Link to WAR-004

WAR-005 Queued Heroic Strike/Cleave hit table Resolved · high

Question: Does queueing suppress the off-hand dual-wield miss penalty at level 60?

Source-backed answer · moderate confidence: The October 1 beta notes state that queueing Heroic Strike no longer increases off-hand hit chance. The engine already models this: the queue aura never removes the off-hand dual-wield penalty, and only the replacing main-hand attack rolls on the special-attack table.

Current implementation / scenario: Attacks remain queued replacements; the penalty is removed only while a queue is active. Low-level observations support the effect, not its full boss-level magnitude.

Impact: Fury hit requirements, queue timing and resource optimization.

Remaining scope: None for the simulated cases; level-63 miss rates belong to CORE-001.

Implementation / model follow-up: Already modelled (sim/warrior/heroic_strike_cleave.go); no change.

Optional / conditional follow-up

Current beta · level 30
Requires: Level-20 dual wield, known weapon skill and isolated queued/unqueued periods.
Observe / verify: Optional cancellation test: cancel before the main-hand swing and measure only off-hand events occurring inside the queue window.
Beyond level 30 / launch
Requires: Level-60 Warrior against a level-63 target.
Observe / verify: Include queued off-hand events in CORE-001's table measurement rather than commissioning a duplicate full test.

sim/warrior/heroic_strike_cleave.go · https://github.com/magey/forever-warrior/issues/2 · docs/upstream-forever-review-2026-09-23.md · https://us.forums.blizzard.com/en/wow/t/wow-classic-forever-beta-patch-notes/2360696

Earlier references: T50 in in_game_checks.md (queue)

Related: CORE-001

Link to WAR-005

WAR-016 Deep Wounds payload and refresh Resolved · high

Question: Which weapon contributes, and is the intended rollover/tick rule live at every rank?

Source-backed answer · moderate confidence: A developer response relayed in Fight Club (2026-10-05) states the intended behaviour: Deep Wounds rolls over its damage when refreshed, does not reset its tick timer, does not scale with Attack Power (an off-hand formula that still used AP was a bug) and scales with weapon damage; the talent connects to Impale again. This is the implemented model.

Current implementation / scenario: Weapon-only MH-average payload, no AP/caster multiplier, carried-over unpaid damage and preserved tick timer; target modifiers apply once.

Impact: All Warrior bleed/talent/APL comparisons.

Remaining scope: Exact payload attribution per hand only.

Implementation / model follow-up: Keep the weapon-only rolling model qualified, remove the obsolete prerequisite-source hold and retain narrow in-game hand/AP/timing controls.

Optional / conditional follow-up

Current beta · level 30
Requires: Known Deep Wounds rank; separated MH/OH weapon damage, AP-only gear change and timed critical strikes.
Observe / verify: Measure bleed increments, next-tick timestamps and remaining-damage rollover; record client/build and triggering hand.
Beyond level 30 / launch
Requires: Higher ranks and final level 60weapon/talent data.
Observe / verify: Confirm rank scaling and any unresolved OH attribution, not a whole new rage model.

docs/october-8-review.md · sim/warrior/deep_wounds.go · https://github.com/ClassicWoWCommunity/forever-bugs/issues/234#issuecomment-6002676309 · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · tools/forever_bench/october9_patch_test.go

Related: DATA-014

Link to WAR-016

WAR-019 Slam swing timer and Improved Slam cooldown Resolved · high

Question: Does untalented Forever Slam interrupt melee swings, and how much cooldown does Improved Slam remove?

Source-backed answer · high confidence: The 'Panel' rule that Slam no longer resets the swing timer is not what the client ships. Untalented ranks (1240193, 1464, 8820, 11604, 11605) keep InterruptFlags 15 (including Combat) and lack SPELL_ATTR2_DO_NOT_RESET_COMBAT_TIMERS, like Era's Slam. Improved Slam (12862) overrides every rank (aura 332) with 1310196–1310200, which have InterruptFlags 7 and set DO_NOT_RESET_COMBAT_TIMERS; its text says 'In addition, Slam no longer interrupts or delays your melee swing'. Its cooldown modifier is −3000 ms at both ranks (curve [0.25, 3], [0.5, 3]); the engine had 1.5 s per point.

Current implementation / scenario: Untalented Slam: no autos during the cast; both swings restart from its completion. Improved Slam: autos continue and no reset. 18-second cooldown, 15 seconds with either rank of Improved Slam.

Impact: All Warrior DPS rotations that cast Slam; the Fury and Arms benchmark rotations.

Remaining scope: None.

Implementation / model follow-up: Implemented in slam.go with a cast-speed-correct reset; regression TestForeverSlamSwingTimerDependsOnImprovedSlam. The benchmark Warrior rotations still cast Slam at 30 (15 for Arms) Rage; with the corrected engine, Fury and Fury (Sunder) gain 4.5%/4.3% without Slam, and Arms/2H Bloodthirst 2.0%/0.7% with a 50-Rage gate.

sim/warrior/slam.go · docs/forever_rules.md · docs/auto_attack_audit.md · https://wago.tools/db2/SpellMisc/csv?build=1.60.1.70338 · https://wago.tools/db2/SpellInterrupts/csv?build=1.60.1.70338

Link to WAR-019

WAR-020 Rend and Thunder Clap Attack Power scaling and Thunder Clap threat Accepted working answer · high

Question: How do Rend and Thunder Clap scale with Attack Power in Forever, and how much threat does Thunder Clap generate?

Credible working answer · moderate confidence: The class deep dive: 'Rend: Now gains significant damage scaling from Attack Power'; 'Thunder Clap: Now gains significant damage scaling from Attack Power ... Threat rebalanced (reduced) to account for being usable in Defensive Stance.' The engine had neither. Fight Club measurements: Rend 0.02 AP per tick at rank 2 using current AP, with Improved Rend scaling the AP term; Thunder Clap about 0.068–0.07 AP, threat equal to damage.

Current implementation / scenario: Rend ticks add 2% of current AP each (Improved Rend multiplies the whole tick). Thunder Clap adds 7% AP to each target's damage and generates threat equal to its damage before stance/talent multipliers.

Impact: Arms (Rend upkeep) and Protection Warrior threat; multi-target Thunder Clap damage.

Remaining scope: Max-rank coefficients.

Implementation / model follow-up: Implemented (Forever only). Arms moves from −5.6% to −2.9% overall after the Slam correction.

Optional / conditional follow-up

Current beta · level 30
Requires: Level-30 Warrior, two gear sets differing only in Attack Power, Improved Rend 0/3.
Observe / verify: Rend tick and Thunder Clap non-crit damage with each set; about 50 ticks and 30 Thunder Claps per set.
Beyond level 30 / launch
Requires: Level-60 Warrior with the top ranks.
Observe / verify: Repeat the comparison for the top ranks.

sim/warrior/rend.go · sim/warrior/thunder_clap.go · https://worldofwarcraft.blizzard.com/en-us/news/24301514/world-of-warcraft-forever-class-deep-dives-priest-and-warrior

Link to WAR-020

WAR-021 Revenge Attack Power scaling and threat Accepted working answer · high

Question: What is Revenge's AP coefficient and threat formula in Forever?

Credible working answer · moderate confidence: A six-point fit at rank 2 in the published 'Warrior Threat Values' sheet (https://docs.google.com/spreadsheets/d/e/2PACX-1vQxd44ah_iUJNcTdH2mpI9nTMNbQEGV9COXkD1nRHi6NGxiU-TvdtkIJBik7KkgN6gKCs7Sn2E-LVxU/pubhtml): threat = 1.495 x 2.25 x (damage roll + 0.28 AP), residuals under 0.5 threat, with no flat component. A separate tester measured 0.20 AP at rank 1 (level 20, 130–252 AP, weapon-independent). The engine previously had no AP term and a Classic-style 2.25 x 2 x level flat bonus.

Current implementation / scenario: Revenge damage = rank roll + 0.28 AP (x Improved Revenge); threat = 2.25 x damage with no flat bonus.

Impact: Protection Warrior damage and threat; the largest single change in this audit (Protection DPS +34% at level 60).

Remaining scope: The rank-6 coefficient: the two low-rank measurements differ (0.20, 0.28), so the level-60 value may be higher or lower. Treat Protection results as sensitive to this value.

Implementation / model follow-up: Implemented with 0.28 for rank 6 (Forever only).

Optional / conditional follow-up

Current beta · level 30
Requires: Level-30 Warrior (Revenge rank 2), two gear sets differing only in Attack Power.
Observe / verify: Revenge non-crit damage and threat; about 50 casts per set.
Beyond level 30 / launch
Requires: Level-60 Warrior (Revenge rank 6).
Observe / verify: Repeat to replace the extrapolated 0.28 AP coefficient.

sim/warrior/revenge.go

Link to WAR-021

WAR-007 Unbridled Wrath beta fix status Resolved · medium

Question: Has the acknowledged lower-than-intended beta proc rate been corrected?

Source-backed answer · high confidence: Public foreverlogs.gg reports show Unbridled Wrath rage gains on 56.7-59.0% of eligible white swings for four level-30 Fury Warriors (2,008 of 3,492 in total), which matches the fixed 12% per point (60% at 5/5). A fifth Warrior at 34% fits 3/5.

Current implementation / scenario: Intended 12% per point on ordinary landed white swings, granting one Rage with either weapon layout. HS/Cleave excluded; the calculator no longer advertises two Rage for two-handed weapons.

Impact: Warrior rage supply; intended model may differ from a live beta bug.

Remaining scope: Whether extra swings from other procs are eligible.

Implementation / model follow-up: Already modelled; no change.

Optional / conditional follow-up

Current beta · level 30
Requires: Only if matching a specific live beta build is necessary; known UW rank and ordinary landed autos.
Observe / verify: Count energize events, including capped zero-gain events, against eligible swings. Record the build; do not count HS/Cleave as white swings.

docs/flurry_review.md · sim/warrior/talents.go · https://github.com/ClassicWoWCommunity/forever-bugs/issues/105 · docs/protection_sunder.md · docs/october-8-review.md · https://foreverlogs.gg/

Earlier references: T58

Link to WAR-007

WAR-009 Sunder Armor AP threat coefficient Accepted working answer · medium

Question: What AP-scaled threat was added by the patch beyond the explicit rank threat?

Credible working answer · moderate confidence: The Priest and Warrior class deep dive states 'Sunder Armor: Threat generated now scales with Attack Power'. A published 'Warrior Threat Values' test sheet (https://docs.google.com/spreadsheets/d/e/2PACX-1vQxd44ah_iUJNcTdH2mpI9nTMNbQEGV9COXkD1nRHi6NGxiU-TvdtkIJBik7KkgN6gKCs7Sn2E-LVxU/pubhtml) solves 0.0433 AP at rank 1 and 0.0476 at rank 2 on top of the client's flat 34 and 75; he notes the coefficient may rise with rank. 0.05 is the provisional level-60 value.

Current implementation / scenario: Forever Sunder threat is the client's flat effect (rank 5: 206) plus 0.05 x Attack Power, before stance and talent multipliers.

Impact: Protection and Fury Sunder threat, not Sunder armor reduction or direct DPS.

Remaining scope: Rank-5 coefficient.

Implementation / model follow-up: Implemented in sunder_armor.go.

Optional / conditional follow-up

Current beta · level 30
Requires: Same Sunder rank, stance/talents and enemy; two known AP values and precise threat API readings.
Observe / verify: Compare threat slopes over multiple applications, separating first application and refresh. No damage-based proxy is needed.
Beyond level 30 / launch
Requires: Rank-5 Sunder if low-rank results establish a nonzero slope.
Observe / verify: Verify whether the slope is rank-independent before applying it at level 60.

sim/warrior/sunder_armor.go · docs/forever-70009.md

Earlier references: 70009 unexposed Sunder threat coefficient

Link to WAR-009

WAR-011 Overpower/Revenge cooldown and Focused Rage Resolved · medium

Question: Are the earlier independent-timer and undiscounted Demoralizing Shout findings still current?

Previously resolved · high confidence: Current registration passes the same timer to Overpower and Revenge. Demoralizing Shout has the offensive flag read by Focused Rage. The old report does not describe a current missing fix.

Current implementation / scenario: No. Warrior.Initialize passes one timer to Overpower and Revenge. Demoralizing Shout carries SpellFlagOffensive, so Focused Rage discounts it.

Impact: The old follow-up report must not create duplicate fixes or false candidate gains.

Remaining scope: None for these two implementation claims; missing Shield Bash is separate.

Implementation / model follow-up: Keep the resolved record rather than repeat the change or schedule a gameplay experiment.

docs/upstream-forever-followup-2026-09-23.md · sim/warrior/warrior.go · sim/warrior/revenge.go · sim/warrior/overpower.go · sim/warrior/demoralizing_shout.go · sim/warrior/talents.go

Related: WAR-012

Link to WAR-011

WAR-012 Shield Bash registration Outside current scenarios · medium

Question: Can tanks use the sourced shield interrupt without bypassing Pummel's cooldown category?

Outside current scenarios · high confidence: Interrupts do not affect damage against the non-casting reference boss. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Pummel is registered; Shield Bash is absent from Warrior.Initialize. The old report claiming two live independent timers overstated current coverage.

Impact: Protection interrupt/utility completeness, not damage in a noncasting reference boss.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Current Bash/Pummel rank and cooldown-category records.
Observe / verify: Check shield and stance restrictions and both cast orders against the shared timer; compare registered damage independently from interrupt behavior.

sim/warrior/warrior.go · sim/warrior/pummel.go · docs/upstream-forever-followup-2026-09-23.md

Related: WAR-008

Link to WAR-012

WAR-015 Gore Drinker heal charges and refresh Outside current scenarios · medium

Question: Which melee events consume the next three healing attacks, and how do repeated triggers refresh the effect?

Outside current scenarios · moderate confidence: Self-healing does not change DPS or threat results; no benchmark build takes Gore Drinker. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Current talent text gives 0.5/1% maximum Health on the next three melee attacks after Enrage, Berserker Rage, Bloodrage, Death Wish or Bloodthirst. The tree marks Gore Drinker not simulated; no guessed healing is supplied.

Impact: Potential Warrior healing/survival and hybrid comparisons. Current migrated benchmark allocations do not take Gore Drinker.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Current hotfix rows for the triggered aura and charge/proc attributes, or a credible event-level Forever study.
Observe / verify: Trace the exact heal trigger, charge decrement, lifetime and refresh rule; separate visible percentage from server scripting.
Beyond level 30 / launch
Requires: At least level35 for the first Gore Drinker point (25 previous Fury points), controlled missing Health, melee outcomes and the relevant trigger abilities. Outside the level30 cap.
Observe / verify: Isolate white/special/queued/extra swings, avoided attacks and multi-target attacks; compare Health gains and charge changes. Trigger again with charges remaining to determine refresh behavior.

https://www.wowhead.com/forever/spell=1323967/gore-drinker · docs/october-1-corrections.md · ui/core/talents/trees/warrior.json · sim/warrior/talents.go

Related: WAR-008

Link to WAR-015

WAR-018 Demoralizing Shout and Roar flat threat Accepted working answer · medium

Question: What exact per-target flat threat do the fixed Forever Shout and Roar generate?

Scenario decision · moderate confidence: The official notes resolve nonzero threat on all affected targets, but publish no numeric amount. Scoped three-target, miss, zero/manual override and Classic-retention checks validate the exposed model, not its server coefficients. October 11 audit: The client now carries an explicit threat effect (SpellEffect 63) on every Demoralizing Shout rank: 11/19/27/35/43, so rank 5 is 43, matching the retained 43.2 convention. A published Fight Club threat sheet measured rank 2 at about 22.7 per target before modifiers on a level-30 character, slightly above the client's 19, suggesting mild level growth.

Current implementation / scenario: Nonzero affected-target behavior is implemented. Maximum-rank defaults are 43.2 Shout and 100 Roar, before stance/global multipliers, as Classic-derived conventions. Player optional override preserves explicit zero; Classic ignores it.

Impact: Warrior/Bear multi-target threat and cost-constrained maintenance.

Remaining scope: Actual per-rank flat threat and any hidden target/distribution behavior beyond the tested model.

Implementation / model follow-up: Record defaults/override with comparisons; do not call 43.2/100 measured values or restore the former zero as intended behavior.

Optional / conditional follow-up

Current beta · level 30
Requires: Level-30 Warrior or Bear with learned Shout/Roar, one and multiple controlled enemies, threat API and no other damage/resource gains.
Observe / verify: Measure each landed target and miss boundary separately; remove stance/global modifiers before estimating flat threat.

https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/warrior/demoralizing_shout.go · sim/druid/demoralizing_roar.go · tools/forever_bench/october9_patch_test.go

Related: DRU-005 · WAR-011 · SCEN-003

Link to WAR-018

WAR-023 Benchmark Warrior rotations after the Slam and Rend corrections Resolved · medium

Question: Which Slam and Rend usage is best for each Warrior benchmark row on the corrected engine?

Source-backed answer · high confidence: Screened Slam at no use and 30–70 Rage, with and without Rend, on every race (seed 20261012), then confirmed on an independent seed (20261099): Fury +5.31% (every race +5.15% or more), Fury (Sunder) +4.98%, Arms +1.93%, 2H Bloodthirst +0.90% (every race positive). Dual-wield Slam loses at every threshold under the swing reset; Fury rows had Improved Rend talented without casting Rend, which now scales with AP. The non-benchmark AoE and Spearing dual-wield presets also drop Slam.

Current implementation / scenario: Fury and Fury (Sunder): no Slam; Rend maintained after Whirlwind. Arms: Slam at 50+ Rage. 2H Bloodthirst: Slam at 50+ Rage and Rend after Whirlwind.

Impact: All four Warrior DPS rows.

Remaining scope: None.

Implementation / model follow-up: Presets and the 40 ranked profiles updated.

ui/warrior/apls/forever_fury.apl.json · ui/warrior/apls/forever_arms.apl.json · ui/warrior/apls/forever_fury_2h.apl.json · ui/warrior/apls/forever_fury_sunder.apl.json

Link to WAR-023

WAR-017 Last Stand temporary-health expiry floor Outside current scenarios · low

Question: How does Last Stand remove its temporary health, particularly near death?

Outside current scenarios · moderate confidence: Affects only near-death survival, which the tank checks do not reach. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Snapshot30% final maximum-health grant, raw-factor compensated; expiration removes actual temporary max/current grant without fake damage/TMI. Dead stays0; living floor1HP is a Classic convention.

Impact: Warrior temporary-health survival/accounting and health-multiplier correctness.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Current beta · level 30
Requires: Access to Last Stand through legal level 30 allocation/Legacy prerequisite context, precise current/max health, near-death expiry.
Observe / verify: Record grant and expiry separately; verify source-backed loss and the unmeasured living floor without conflating incoming damage.

https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/warrior/talents.go · sim/core/health.go · tools/forever_bench/october9_patch_test.go

Related: SCEN-001

Link to WAR-017

WAR-022 Battle Shout threat Resolved · low

Question: Does Battle Shout generate threat in Forever?

Source-backed answer · high confidence: Buffs and debuffs generating 0 threat was confirmed as Not A Bug (forever-bugs#64, 2026-09-22), and the published Fight Club threat sheet measured Battle Shout at 0. Demoralizing Shout's threat was later restored explicitly (WAR-018); Battle Shout's was not.

Current implementation / scenario: No threat in Forever; Classic keeps its per-target flat threat.

Impact: Protection Warrior threat (about 2 TPS at level 60).

Remaining scope: None.

Implementation / model follow-up: Battle Shout FlatThreatBonus 0 under Forever.

sim/warrior/shouts.go · https://github.com/ClassicWoWCommunity/forever-bugs/issues/64

Link to WAR-022

WAR-024 Warrior changes in 1.60.1.70425 Resolved · low

Question: Which Warrior changes affect the simulation?

Source-backed answer · high confidence: Notes; the client changes Lingering Rage to a rage-decay aura (1323965 aura 94 to 110) and rewords Bloodthrill, Gore Drinker and Victorious without changing values.

Current implementation / scenario: Weaponmaster's extra attack no longer triggers from Shield Slam. Battle Shout reaches the whole raid. Slam pushback, Overpower against low-level enemies, Lingering Rage health regeneration and Improved Berserker Rage snare removal have no simulated effect. Human, Orc, Dwarf and Tauren racials changed (CORE-020).

Impact: Protection and Fury tanks with Weaponmaster; racial rows.

Remaining scope: None.

Implementation / model follow-up: sim/warrior/talents.go.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · https://wago.tools/builds-diff?to=1.60.1.70425&from=1.60.1.70374

Link to WAR-024

Paladin 6 active / 14 total

PAL-002 Seal Echo coexistence and consumption Needs evidence · high

Question: Can separate Command/Righteousness Echoes coexist, and what exactly spends them?

Partly answered · moderate confidence: The current Echo aura rows explicitly have one charge and white-melee proc mask 4. That supports the charge count and excludes ordinary Judgements from the consumption family. Separate spell IDs do not prove simultaneous coexistence or what avoided events do. Twist of Light is a level-40 talent, so this is not a level-30 experiment.

Current implementation / scenario: Separate one-charge Echoes wait for a landed white swing. Avoided swings and melee-classified Judgements do not consume them.

Impact: Ret seal-weaving damage; the talent itself needs later access.

Remaining scope: Echo coexistence, replacement and avoided-swing consumption.

Implementation / model follow-up: Retain the sourced one-charge/white-event structure while leaving those script-level details open.

Answer routes

Beyond level 30 / launch
Requires: Level-40+ Twist of Light, both relevant seals and aura/event logging.
Observe / verify: Replace seals before the next auto, record simultaneous Echo auras, then compare landed and avoided autos plus a Judgement-only control.

sim/paladin/twist_of_light.go · docs/upstream-forever-review-2026-09-23.md · https://wago.tools/db2/SpellAuraOptions?build=1.60.1.69977 · ui/core/talents/trees/paladin.json

Earlier references: T53

Link to PAL-002

PAL-003 Windfury Totem proc and charge rules Needs evidence · high

Question: Which specials/seal procs trigger Totem, how are AP charges spent, and do extra attacks reset ordinary swings?

Partly answered · moderate confidence: The sourced amount and separate Totem-versus-Weapon identity do not answer eligible Seal events, AP-charge consumption or extra-swing timing. The existing matched sim audit explains its output but is not an in-game experiment. Totem is learned at 32; level-30 Windfury Weapon is a different effect and cannot substitute for the test.

Current implementation / scenario: Separate party buff, exclusive with Grace; not a weapon imbue. Retained ICD, extra-swing timing and seal eligibility remain inherited. There is no second 20% roll to spend a charge.

Impact: All melee and tank rows; Ret's autoattack gain can exceed the displayed 20% proc chance because specials also trigger swings.

Remaining scope: Totem proc-event eligibility, ICD, charge use and swing timing.

Implementation / model follow-up: Preserve the separate air buff and its exclusivity. Treat old matched DPS numbers as an implementation explanation, not a source for the server proc model.

Answer routes

Beyond level 30 / launch
Requires: Level-32+ Shaman providing Totem, Paladin recipient, no weapon imbue or other extra-attack source.
Observe / verify: Count eligible autos, specials and seal events; record AP aura charges and the next normal swing. Isolate avoided events and exclude immediate recursion before estimating chance.

sim/core/buffs.go · docs/windfury.md · docs/upstream_elliot_review.md · docs/forever-70009.md · https://wago.tools/db2/SpellLevels?build=1.60.1.69977

Earlier references: T28 / T30 (Windfury interactions)

Link to PAL-003

PAL-006 Seal of Fury shield replacement and mana Needs evidence · high

Question: Does a new proc replace or add to the shield, and which depletion events restore mana?

Partly answered · moderate confidence: The talent-only/full-absorption/attacker-level contract is sourced and tested. Expiry, replacement and cancellation do not refund. Replacement rather than accumulation and an independently timed30-second pool are approved assumptions, not established by the seal duration.

Current implementation / scenario: One shared replacement shield across ranks; 50% actual dealt Holy proc damage with a shield equipped. Configurable lifetime absent30s/explicit0 no absorption. Only talented actual full depletion returns level-based mana from the incoming attacker, with NoThreat.

Impact: Protection survival and resource sustain; must not grant free mana on expiry or replacement without support.

Remaining scope: Actual server replacement/addition, shield lifetime and simultaneous-event ordering. No accumulation mode or judgement shield is claimed.

Implementation / model follow-up: Record lifetime in scenario inputs; compare no-absorb/short/30-second controls. Do not promote a working convention to resolved script behavior.

Answer routes

Current beta · level 30
Requires: Level 20+ Paladin with Improved Seal of Fury, shield, learned ranks, uncapped mana and known incoming attacker levels.
Observe / verify: Measure successive procs, rank switching, partial/full depletion, replacement/expiry and capped attacker-level returns; distinguish generated capacity, absorbed damage and mana.

docs/mythicsim_review.md · https://www.wowhead.com/forever/spell=20423/seal-of-fury · https://www.wowhead.com/forever/spell=1314104/improved-seal-of-fury · https://www.wowhead.com/forever/spell=1314103/improved-seal-of-fury · https://wago.tools/db2/Spell?build=1.60.1.69977 · https://wago.tools/db2/SpellEffect?build=1.60.1.70009 · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/paladin/sof.go · proto/paladin.proto · tools/forever_bench/sof_model_october9_test.go · sim/core/shield_depletion_test.go

Earlier references: T45 shield portion / M02

Related: CORE-014 · SCEN-005

Link to PAL-006

PAL-008 Vindication's target AP reduction Needs evidence · high

Question: Does the target-side part of Vindication reduce incoming damage?

Partly answered · moderate confidence: Vindication applies sourced level-60 AP reduction 68/136/204 with strongest-effect handling while preserving the owner’s separate damage bonus. Incoming-damage native controls pass.

Current implementation / scenario: Vindication applies sourced level-60 AP reduction 68/136/204 with strongest-effect handling while preserving the owner’s separate damage bonus. Incoming-damage native controls pass.

Impact: Tank mitigation and fair talent selection.

Remaining scope: Lower-level evaluation/rounding remains qualified rather than proved by the max-level test.

Implementation / model follow-up: Corrected locally with focused native controls; affected exact profiles and published results require refresh before these changes are released.

Answer routes

Sources / code
Requires: Trait curve 82984 and child 440667 level scaling plus existing AP-debuff exclusivity.
Observe / verify: Reconcile the rendered level-60 amount with raw level fields; test activation/expiry and competing debuffs against known target AP.
Current beta · level 30
Requires: Paladin with 1/3 to 3/3 Vindication (10 Retribution points first) and matched enemy swings.
Observe / verify: Validate the target aura and damage reduction at each rank, without confusing owner AP growth with enemy AP loss.

sim/paladin/talents.go · sim/core/attack.go · docs/beta-pass/paladin.md · https://www.wowhead.com/forever/spell=9452/vindication · https://www.wowhead.com/forever/spell=440667/vindication · tools/forever_tanks/main.go

Link to PAL-008

PAL-004 Holy melee partial resistance Needs evidence · medium

Question: Do melee-table Holy strikes and seal damage take partial resists?

Game test needed · low confidence: Holy school plus a melee hit table is not a physical-school spell, and the current code intentionally retains the resistance path. No controlled Forever report located in this pass demonstrates an exception for Holy Strike or the seal proc. Their resistance treatment cannot be inferred from the melee classification alone.

Current implementation / scenario: Current melee hit/crit handling retains the applicable resistance path; table classification alone does not settle partial resistance.

Impact: Paladin damage and threat against higher-level targets.

Remaining scope: Partial-resist eligibility, separate from avoidance and crit.

Implementation / model follow-up: Do not add IgnoreResists simply to improve Paladin damage.

Answer routes

Current beta · level 30
Requires: Holy Strike and seal attacks against targets of known equal and higher levels; fixed damage inputs.
Observe / verify: Inspect explicit resisted amounts and damage clusters separately from miss/dodge outcomes. A small sample with no resist is not proof of immunity.

sim/paladin/holy_strike.go · sim/paladin/sor.go · docs/beta-pass/paladin.md · sim/core/spell_resistances.go

Earlier references: T28 (Holy resist portion)

Link to PAL-004

PAL-010 Retribution Aura provider scaling Needs evidence · medium

Question: How should the aura use its provider's SP rather than the recipient's?

Partly answered · moderate confidence: October 8 publishes the six-percent coefficient and Vanilla base restoration, superseding the old Forever30/unsourced-coefficient answer. Provider ownership is represented explicitly and scoped dynamic/party controls pass.

Current implementation / scenario: Registered maximum Thorns18/Retribution Aura20 plus0.06 provider school damage power. Owned provider references are explicit/dynamic; declared external SP defaults0. No recipient-SP fallback. Classic behavior remains separate.

Impact: Tanks, solo damage and multi-player buff modeling.

Remaining scope: Reflection outcome/threat attribution and configured coverage/range conventions; low-rank behavior beyond registered maximum support.

Implementation / model follow-up: Keep owned/external metadata separate and record actual provider assumptions in comparisons; no automatic inferred caster or recipient-SP substitution.

Answer routes

Current beta · level 30
Requires: Paladin level 16+ or Druid Thorns, distinct caster/recipient, fixed enemy and two caster-SP controls.
Observe / verify: Measure base/SP and change caster gear after application; hold recipient gear fixed. Check party/range and damage/threat ownership separately.

sim/core/buffs.go · docs/upstream-forever-review-2026-09-23.md · docs/forever-70009.md · https://github.com/ClassicWoWCommunity/forever-bugs/issues/117#issuecomment-5822983107 · https://wago.tools/db2/SpellEffect?build=1.60.1.70009 · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · tools/forever_bench/reflection_october9_test.go

Related: CORE-008 · DRU-011

Link to PAL-010

PAL-001 Seal of Righteousness spell-power formula Accepted working answer · high

Question: Should the proc add SP once or twice, and is the retained two-hand 1.1 factor real in Forever?

Credible working answer · moderate confidence: The October 15 notes: 'Fixed an issue causing Seal of Righteousness to gain almost twice as much benefit from Spell Damage than intended.' Client 1.60.1.70425 also zeroes the aura's own coefficient (25713-25742 e0 0.1 to 0). The double addition was the in-game behavior until now and is gone.

Current implementation / scenario: The double addition is removed: the proc adds its spell-power contribution once, through its coefficient. The two-hand factor is retained.

Impact: Both Ret gear archetypes and Protection damage/threat; high-impact scaling uncertainty.

Remaining scope: The two-hand factor and the target Holy-bonus treatment.

Implementation / model follow-up: sim/paladin/sor.go (the proc keeps its 0.1 coefficient, applied once).

Optional / conditional follow-up

Current beta · level 30
Requires: Same low-rank seal, target and recorded weapon speed; two known personal-SP values and one-/two-hand controls.
Observe / verify: Measure proc damage slopes without Judgement of the Crusader, then add that target bonus separately. Keep weapon damage, talents and crit outcomes controlled.

sim/paladin/sor.go · docs/weekly_review.md · docs/upstream_elliot_review.md · https://us.forums.blizzard.com/en/wow/t/sor-needs-the-tbc-formula/2355721 · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · https://wago.tools/builds-diff?to=1.60.1.70425&from=1.60.1.70374

Earlier references: T57 (Righteousness)

Link to PAL-001

PAL-005 Seal of Fury and its judgement Accepted working answer · high

Question: What supported Seal of Fury functionality is still missing from Protection?

Partly answered · high confidence: The former blanket missing-family claim is superseded by source-backed rank registration and scoped native checks. Damage/shield/talent-only depletion mana are represented; judgement has its sourced four-second taunt aura, not a complete aggro/target-selection engine.

Current implementation / scenario: Seven Seal of Fury ranks, white-hit damage, shield requirement, judgement and one-white-charge Echo are implemented locally. One shared replacement pool and configurable default 30-second shield lifetime are explicitly provisional; canonical profiles are not switched to SoF.

Impact: Major Protection completeness gap; may change gear, rotation, survival and threat.

Remaining scope: Shield script conventions under PAL-006; unproven judgement absorb/cap-shaped dummy fields and missing aggro semantics are not invented. Final all3-race fixed-lifetime comparisons reject SoF promotion; source defaults retain SoR/oil/Swift/HolyShield. Deployment is separately evidenced.

Implementation / model follow-up: Retain supported rank/cost/SP and no-shield/miss/special boundaries. Do not label configured replacement/lifetime or a taunt aura as fully verified server behavior.

Optional / conditional follow-up

Sources / code
Requires: Current parent/child rank chain and finite-absorb infrastructure.
Observe / verify: Implement legal selection, rank/cost mapping, damage, judgement and shield-only absorb; test no-shield and iteration-reset behavior.
Current beta · level 30
Requires: Level-10/18 Seal of Fury and a shield.
Observe / verify: Validate the implemented low-rank proc damage and absorb relationship; use PAL-006 for replacement/depletion behavior.

docs/mythicsim_review.md · docs/tank_benchmark.md · sim/paladin/paladin.go · https://wago.tools/db2/Spell?build=1.60.1.69977 · https://wago.tools/db2/SpellEffect?build=1.60.1.70009 · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/paladin/sof.go · tools/forever_bench/sof_october9_test.go · tools/forever_bench/sof_model_october9_test.go

Earlier references: T45 / M01

Related: PAL-006 · SCEN-012

Link to PAL-005

PAL-007 Hammer of the Righteous availability Outside current scenarios · medium

Question: Is the retained spell actually learnable, and what shares its cooldown?

Outside current scenarios · low confidence: Hammer of the Righteous helps only against several targets; the benchmark is single-target. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Not registered. Hammer of the Righteous is a learnable level-40 Forever Paladin ability that shares Holy Strike's cooldown.

Impact: Potential Protection AoE action; absent from current rotations.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Matching-build trainer/taught-spell or live talent evidence for 407632.
Observe / verify: Trace the player acquisition path rather than merely finding a class skill row.
Current beta · level 30
Requires: Trainer/spellbook UI with unavailable future ranks visible, if the beta exposes them.
Observe / verify: Capture an actual offering and requirement. No fixed later test level can be promised until this evidence exists.

docs/mythicsim_review.md · docs/spell_coverage.md · artifacts/spell_coverage.json

Earlier references: T46 / M03

Link to PAL-007

PAL-009 Consecrated Ground per-target coverage Resolved · medium

Question: Does the Holy-damage bonus apply only against the first four enemies covered by Consecration?

Source-backed answer · high confidence: Consecrated Ground amplifies owned Holy damage on the first four stationary targets, not every encounter target and not physical damage. Four-target native scope checks pass. Moving-ground geometry remains outside the fixed reference scenario.

Current implementation / scenario: Consecrated Ground amplifies owned Holy damage on the first four stationary targets, not every encounter target and not physical damage. Four-target native scope checks pass. Moving-ground geometry remains outside the fixed reference scenario.

Impact: Can amplify Holy damage against uncovered enemies or beyond the four-target cap. The cap does not distinguish the stationary one-/three-target workloads.

Remaining scope: None for the resolved question. Current201-profile results/native-web defaults are refreshed; actual publication is separately evidenced by deploy.yml and live resource checks.

Implementation / model follow-up: Corrected locally with focused native controls; affected exact profiles and published results require refresh before these changes are released.

sim/paladin/talents.go · sim/paladin/consecration.go · docs/beta-pass/paladin.md · https://www.wowhead.com/forever/spell=1310905/consecrated-ground · https://wago.tools/db2/TraitDefinitionEffectPoints?build=1.60.1.69977

Link to PAL-009

PAL-011 Redoubt proc curve Resolved · medium

Question: Does Redoubt proc at 2/4/6/8/10% by rank?

Source-backed answer · moderate confidence: Public foreverlogs.gg reports for six Protection Paladins show Redoubt on 10.0% of isolated landed melee attacks (1,404 of 14,094) and 9.8% of isolated missed, dodged or parried attacks (274 of 2,794), counting only attacks with no other attack within 300 ms. This matches 10% at 5/5 and shows that avoided attacks also roll.

Current implementation / scenario: Any melee attack against the Paladin, landed or avoided, has a 2% chance per rank to grant the Redoubt block bonus (4% block per rank, 10 seconds or 5 blocks).

Impact: Protection block uptime and defensive talent value.

Remaining scope: Proc chance at lower ranks (talent ranks are not in the logs).

Implementation / model follow-up: Implemented: avoided attacks now roll in Forever (sim/paladin/talents.go).

sim/paladin/talents.go · docs/beta-pass/paladin.md · https://wago.tools/db2/TraitDefinitionEffectPoints?build=1.60.1.69977 · https://wago.tools/db2/CurvePoint?build=1.60.1.69977 · https://wago.tools/db2/SpellAuraOptions?build=1.60.1.69977 · ui/core/talents/trees/paladin.json · https://foreverlogs.gg/

Link to PAL-011

PAL-012 Echo of Justice proc probability Outside current scenarios · medium

Question: What is its intended proc probability after the excessive-rate beta fix?

Outside current scenarios · high confidence: The traced Justice seal/echo produces a stun; its judgement prevents fleeing. No direct damage effect is present in that chain, and current reference workloads model neither control nor fleeing. Its exact excessive-proc-rate fix therefore is not a required damage/threat benchmark experiment.

Current implementation / scenario: No demonstrated source-supported implementation is enabled for the unsupported effect.

Impact: Conditional talent/item interactions; not a granted ranking bonus.

Remaining scope: Proc probability and control interactions if a stunnable/fleeing encounter is introduced. Such control can affect survival in a different workload.

Implementation / model follow-up: Keep in the control-capability backlog rather than the current test queue. Do not apply this dismissal to damaging Command/Righteousness Echoes.

docs/forever-70009.md · https://wago.tools/db2/SpellEffect?build=1.60.1.70009 · https://wago.tools/db2/Spell?build=1.60.1.69977

Link to PAL-012

PAL-016 Reckoning charges Accepted working answer · medium

Question: How are Reckoning charges stored and released?

Credible working answer · moderate confidence: Notes and client 20177 ('Stores up to 4 extra attacks. Cannot occur from Blocks more than once every 1.5 sec').

Current implementation / scenario: Each trigger stores a charge (up to 4); the next main-hand auto attack releases all of them as extra attacks. Only the block trigger keeps the 1.5-second cooldown.

Impact: Protection Paladin with Reckoning.

Remaining scope: Whether the released attacks reset the swing timer.

Implementation / model follow-up: sim/paladin/talents.go.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · https://wago.tools/builds-diff?to=1.60.1.70425&from=1.60.1.70374

Link to PAL-016

PAL-013 Light's Vigil damage support Outside current scenarios · low

Question: Can the supported enemy-target Holy Shock interaction be represented?

Outside current scenarios · moderate confidence: No benchmark build takes Light's Vigil. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Light’s Vigil has ranked enemy casts, one owned enemy mark, enemy damage/refund and explicit linked Holy Shock cooldown behavior. Native cooldown/refund controls pass; manual proc casts are rejected.

Impact: Conditional Holy Shock builds and spellbook completeness, not a proven gain for existing Ret.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Vigil parent ranks 1310911/1311590/1311595 and their enemy damage children.
Observe / verify: Test rank/cost mapping, one active mark, one consumption, refund and Holy Shock cooldown interaction, including target changes.
Beyond level 30 / launch
Requires: Level-40+ Light's Vigil and Holy Shock.
Observe / verify: Confirm the linked cast sequence and amount at the first available rank; later verify max-rank scaling.

docs/beta-pass/paladin.md · https://wago.tools/db2/Spell?build=1.60.1.69977 · https://wago.tools/db2/SpellEffect?build=1.60.1.70009 · ui/core/talents/trees/paladin.json · sim/paladin/lights_vigil.go

Link to PAL-013

Druid 9 active / 18 total

DRU-003 Form weapon damage and flat imbues Needs evidence · high

Question: Which equipped weapon damage and flat imbue bonuses affect Cat/Bear form attacks?

Partly answered · moderate confidence: Equipped DPS, stones and striking remain supported in form weapons. October 8 Flametongue controls confirm active Cat1/Bear2.5-second speed rather than equipped speed; no second form damage fix.

Current implementation / scenario: Officially supported equipped weapon DPS, stones and striking enchants now enter form weapons. Cat uses its 1-second weapon and Bear its 2.5-second weapon; native rebuild/swap controls pass without duplicating AP or flat damage.

Impact: Feral/Bear weapons, stones and damage scaling.

Remaining scope: Full flat-imbue script formulas and acquisition remain separate; Flametongue coefficient is qualified under SHA-006.

Implementation / model follow-up: Retain the current active form reconstruction and avoid treating Classic/dead branches as discrepancies.

Answer routes

Current beta · level 30
Requires: Bear and level-20 Cat, matched-stat weapons with substantially different DPS, plus an obtainable flat-damage enchant/imbue.
Observe / verify: Compare form attack damage at fixed AP; test the flat effect independently from the weapon swap.
Beyond level 30 / launch
Requires: Level-60 forms and high-level weapon effects if low-level results imply scaling.
Observe / verify: Verify the actual level-60 base and any effect-specific exceptions before changing synthetic-weapon valuation.

sim/druid/forms.go · sim/core/attack.go · docs/upstream_elliot_review.md · sim/core/item_effects.go · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · tools/forever_bench/druid_hunter_shaman_october9_test.go

Earlier references: T18 / T60

Related: SHA-006

Link to DRU-003

DRU-005 Bear threat multipliers Needs evidence · high

Question: What are Forever's form, Primal Bite, Lacerate and Swipe threat multipliers?

Game test needed · low confidence: October 8 explicitly raises the form bonus from30% to50%; active enter/exit and Classic controls pass. The October 1 relative Primal Bite correction does not certify its absolute coefficient.

Current implementation / scenario: Bear/Dire Bear form threat1.5× in Forever (Classic1.3×); Primal Bite3×, Lacerate3.33× and Swipe2× retain their qualified ability-specific conventions.

Impact: Tank threat coverage and comparisons; these must not be marketed as verified threat formulas.

Remaining scope: Absolute ability-specific threat coefficients and precise Primal Bite scaling; not the now-sourced form multiplier.

Implementation / model follow-up: Keep the form correction separate from uncertain ability multipliers and demoralizing flat threat.

Answer routes

Current beta · level 30
Requires: Bear, white attacks/Swipe, precise threat readings, known damage and no Thorns/proc interference.
Observe / verify: First determine the form multiplier from ordinary hits, then Swipe's additional threat; include multiple enemies without duplicating totals.
Level-30 beta
Requires: Primal Bite talent, available from 25.
Observe / verify: Isolate direct Bite damage and threat with the previously measured form multiplier.
Beyond level 30 / launch
Requires: Lacerate from 42 and eventually level-60 ranks.
Observe / verify: Separate application and bleed threat at each stack; distinguish flat threat from damage multipliers.

sim/druid/forms.go · sim/druid/mangle.go · sim/druid/lacerate.go · sim/druid/swipe.go · docs/tank_benchmark.md · https://us.forums.blizzard.com/en/wow/t/feral-druid-threat/2355903 · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-1/2360696/4 · docs/october-1-corrections.md · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · tools/forever_bench/druid_hunter_shaman_october9_test.go

Earlier references: Bear threat notes in beta-pass/druid.md

Link to DRU-005

DRU-006 Lacerate direct and periodic formula Needs evidence · high

Question: Are the stack-dependent weapon term and bleed base the correct Forever damage formula?

Partly answered · moderate confidence: The parent rows source 10/12/15 bleed damage per three-second tick, five stacks and a 10% weapon-damage term. The uncertainty is narrower than the whole formula: the description says 'per existing application', whereas the engine uses the resulting stack including the new application. Child 414647 has a separate 20% weapon effect and is not proof of its current script use.

Current implementation / scenario: 10% form weapon damage per resulting stack and level-dependent bleed base per stack; inherited interpretations remain unvalidated.

Impact: Bear rotation and damage, distinct from its threat multiplier.

Remaining scope: Pre- versus post-application stack count, exact child dispatch and any hidden AP/script term.

Implementation / model follow-up: Do not automatically double the direct coefficient from an untraced child record. Keep known bleed ranks distinct from the unresolved direct-stack ordering.

Answer routes

Beyond level 30 / launch
Requires: Level-42+ Bear, stable AP, no other bleeds, isolated first through fifth applications.
Observe / verify: Determine whether the first application deals direct damage and whether subsequent hits use the pre-existing or new stack count; separately measure tick damage and crits.

sim/druid/lacerate.go · docs/forever_rules.md · docs/beta-pass/druid.md · https://www.wowhead.com/forever/spell=414644/lacerate · https://wago.tools/db2/SpellEffect?build=1.60.1.70009

Related: DRU-005

Link to DRU-006

DRU-011 Proc enchants in animal forms Needs evidence · high

Question: Can Cat and Bear trigger Crusader, and which weapon speed determines its proc rate?

Partly answered · moderate confidence: October 8 specifically corrects Recovery and Revelation form behavior. Recovery is implemented with source-backed eligibility/payload and scoped checks; Revelation's probability law is not supplied by the note. October 11 audit: The WoW Forever Druid community FAQ (Google Doc linked from the WoW Forever US/NA #druid channel) reports that Cat and Bear weapons can proc on-hit effects in form and that feral damage scales with weapon DPS, which supports form eligibility; it does not give proc frequencies.

Current implementation / scenario: Source-backed Recovery dodge/parry heal and ten-second gate work in forms; existing Crusader/Lifestealing and form stones/oils remain. Revelation form retention is sourced and its explicit provisional proc model is integrated locally, not released.

Impact: Druid enchant selection, temporary Strength and resulting damage/threat. Simulation confirmation does not establish that the proc occurs in game.

Remaining scope: Other proc eligibility/frequencies and Revelation law under DATA-015; form eligibility alone is not a universal proc formula.

Implementation / model follow-up: Keep exact equipped enchant source chains and qualified rates; no wearer profession requirement is inferred from a recipe.

Answer routes

Current beta · level 30
Requires: Only if an enchanted weapon can actually be supplied in the beta; an obtainable enchant is not established by a low character-level requirement.
Observe / verify: Compare isolated caster-form and Cat/Bear attacks with proc timestamps and equipped/form weapon speeds.
Beyond level 30 / launch
Requires: Crusader access if unavailable in either beta cap.
Observe / verify: Repeat the form comparison with enough eligible attacks to distinguish no procs from a low rate. Move earlier if vendors or copied items make the enchant obtainable.

sim/common/enchant_effects.go · sim/druid/forms.go · docs/tank_selection.md · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · tools/forever_bench/enchant_october9_test.go

Related: DRU-003 · DATA-006 · DATA-015

Link to DRU-011

DRU-012 Bear outgoing and incoming rage formulas Needs evidence · high

Question: Does Forever Bear retain damage-based rage or use normalized rules, and how are avoided swings handled?

Partly answered · low confidence: October 8 supersedes October 1's +75% outgoing critical bonus with +100%. This does not establish Warrior-style Bear base outgoing normalization. Incoming semantics are sourced, but the selected numerical scenario is explicitly provisional.

Current implementation / scenario: Bear base outgoing/avoidance interpretation remains inherited; outgoing white critical Rage is now +100%, with Maul excluded. Incoming Rage shares the explicit configurable Forever model under WAR-003; Classic remains unchanged.

Impact: Bear Maul uptime, other ability throughput, rage from boss damage and gear/talent value. Frequent Maul alone cannot validate the model.

Remaining scope: Base outgoing damage/speed/avoidance dependence and level 60 extrapolation; incoming formula terms under WAR-003.

Implementation / model follow-up: Preserve the new critical ratio and disclose the shared incoming assumptions without claiming identical measured class scripts.

Answer routes

Current beta · level 30
Requires: Bear, uncapped rage, no Maul/Enrage/Primal Fury and matched attack speed while damage varies.
Observe / verify: Separate ordinary hits, crits and avoided swings. For incoming rage stop autos and vary armor independently from maximum health.
Beyond level 30 / launch
Requires: Level-60 Bear and known outgoing/incoming attacks.
Observe / verify: Check that the low-level formula still fits; test queued Maul separately because it replaces the ordinary white event.

sim/core/rage.go · sim/druid/tank/tank.go · sim/druid/maul.go · docs/protection_queue.md · https://us.forums.blizzard.com/en/wow/t/bear-druids/2358994?page=2 · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-1/2360696/4 · docs/october-1-corrections.md · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/core/forever_incoming_rage.go · tools/forever_bench/october1_patch_test.go

Related: WAR-001 · WAR-003 · DRU-005 · SCEN-001

Link to DRU-012

DRU-017 Rake Attack Power scaling Needs evidence · high

Question: How much Attack Power does Rake's hit and bleed gain?

Credible working answer · moderate confidence: The September 30 class deep dive says Rake now gains increased damage from Attack Power. The 1.60.1.70425 client carries no coefficient (the term is server-side, as with Swipe's 3%). Fitted to public beta logs from October 2-10 (1,115 rank-1 Rake events): bleed ticks of three Cats with known gear (Cat AP about 250, 330-370 and 465) are 24-28, 36 and 42 against a base of 16 with Savage Fury and Rend and Tear, giving 4-5% per tick; the hit, after armor, fits about 1.45 times the tick coefficient, as the base values do (23 against 16).

Current implementation / scenario: Hit +6.5% of Attack Power and each of the three ticks +4.5%, snapshotted on application. With it, Feral keeps Rake up (+6.1%); at two thirds of these values the gain is still +3.1%, and without the term Rake loses 2.9%.

Impact: Feral rotation and talents.

Remaining scope: A direct measurement at two Attack Power values.

Implementation / model follow-up: Forever-only Attack Power terms on the Rake hit and bleed.

Answer routes

Current beta · level 30
Requires: A level-30 Cat with Rake; a way to change Attack Power by a known amount (Battle Shout, a Blessing of Might or an AP item).
Observe / verify: Rake tick damage (non-crit) with and without the Attack Power change, on a target without bleed modifiers.

sim/druid/rake.go · https://news.blizzard.com/en-us/article/24301515/world-of-warcraft-forever-class-deep-dives-hunter-and-druid

Link to DRU-017

DRU-008 Nature's Grace GCD stacking and timing Needs evidence · medium

Question: Does its explicit GCD reduction stack with cast-speed-based GCD reduction, and when does Wrath trigger it?

Partly answered · moderate confidence: Both effects are explicit in the current text: increased spellcasting speed and reduced global cooldown. Thus the presence of two modifiers is not itself evidence of accidental duplication. The unresolved part is their composition and Wrath's proc timing. The active talent first becomes reachable at 30.

Current implementation / scenario: Applies both explicit -10% GCD and +10% cast speed; Wrath retains the inherited completion-time proc timing.

Impact: Balance cast cadence, instant-spell GCD and buff uptime.

Remaining scope: Composition with ordinary cast-speed GCD scaling and impact versus completion timing.

Implementation / model follow-up: Keep both sourced components pending the narrow timing test; do not remove one solely because both influence GCD.

Answer routes

Level-30 beta
Requires: Nature's Grace, cooldown timestamps and controllable Wrath travel distance.
Observe / verify: Compare GCDs with the aura and an isolated cast-speed control; compare aura activation with cast completion and projectile impact.

sim/druid/talents.go · docs/weekly_review.md · docs/history_review.md · https://wago.tools/db2/Spell?build=1.60.1.69977 · ui/core/talents/trees/druid.json

Earlier references: T41

Link to DRU-008

DRU-010 Feral stealth openers Implementation gap · medium

Question: What is needed to support Prowl, Ravage and Pounce?

Known implementation work · high confidence: This is missing supported spellbook functionality, not an unanswered coefficient for the existing rotation. Prowl is available at 20, but Ravage first at 32 and Pounce at 36. The client specifies stealth/behind requirements, damage effects and combo-point awards.

Current implementation / scenario: No opening stealth state or those attacks. Their contribution is absent rather than estimated.

Impact: Feral opener completeness, especially short fights.

Remaining scope: Implementation of one-time stealth state and supported opener effects; high-level runtime checks.

Implementation / model follow-up: Add a source-based implementation task, not a level-20 request to demonstrate unavailable attacks. Opener-only contribution does not imply irrelevance.

Answer routes

Sources / code
Requires: Current learned ranks/effects and an explicit precombat stealth state.
Observe / verify: Implement legal form/position/stealth gates, combo points and one-use opener behavior; ensure normal combat cannot repeat stealth attacks.
Beyond level 30 / launch
Requires: Level-32 Ravage and level-36 Pounce.
Observe / verify: Confirm opener damage and legal states against the implemented ranks.

docs/spell_coverage.md · sim/druid/druid.go · artifacts/spell_coverage.json · https://wago.tools/db2/SpellLevels?build=1.60.1.69977

Earlier references: T51 in check_dispositions.md (stealth backlog)

Link to DRU-010

DRU-018 On-hit effects at the Cat and Bear form attack speed Needs evidence · medium

Question: Do proc-per-minute effects on special attacks in Cat Form use the form's 1.0 attack speed or the equipped weapon's speed?

Credible working answer · moderate confidence: Druid testers confirmed on September 26 that weapon chance-on-hit effects proc in Cat Form; an October 8 question about whether 1.0 speed raises the proc rate of slow weapons went unanswered. Public beta logs (October 2-10) show Flametongue Totem damage on Cat hits at the 1.0 speed: a Wind Spirit Staff (3.3) Druid's Flametongue Attack hits for 18-19 in caster form and 5-6 in Cat Form, and a Manual Crowd Pummeler (2.0) Cat's 642 non-crit Flametongue Attacks are 9-11 with none near the 2.0-speed value, including those from special attacks. No log carries a Cat PPM proc, so the PPM rule is inferred from that. Before this change, 15 proc-trigger effects used the weapon speed on special attacks while Crusader, Fiery Weapon and the 53 weapon-proc helpers used 1.0.

Current implementation / scenario: In Forever the form attack is the equipped weapon's damage per second at the form's speed (Cat 1.0, Bear 2.5). White and special hits both roll PPM effects at the form speed, and speed-scaled on-hit damage uses it too. Classic keeps its tested rule (special attacks use the equipped weapon's speed).

Impact: Feral and Bear with chance-on-hit weapons or proc enchants (not the modeled ranking gear, which has no proc effects).

Remaining scope: Measured PPM proc counts per Cat special attack with a slow weapon.

Implementation / model follow-up: Forever PPM rolls use the form speed for every path.

Answer routes

Current beta · level 30
Requires: A Cat with a slow (3.0+ speed) weapon carrying a PPM effect, such as Crusader, Fiery Weapon or a chance-on-hit weapon.
Observe / verify: Procs per landed white hit and per landed Shred or Claw over a few hundred of each; compare with PPM x 1.0 / 60 and PPM x weapon speed / 60.

sim/core/aura_helpers.go · sim/core/flametongue_totem.go · sim/druid/forms.go

Link to DRU-018

DRU-001 Omen of Clarity proc frequency Accepted working answer · high

Question: Is the client's 100% field a literal proc guarantee after the ten-second ICD?

Credible working answer · moderate confidence: Public foreverlogs.gg reports (691 Druid proc windows: Cat, Bear and caster) confirm the 10-second cooldown and show that the client's 100% is not literal. A two-rate fit after the cooldown gives white hits 5.5-7.5% in Cat Form (1.0-second attacks) and 14.5-18.5% in Bear Form (2.5-second attacks), the same 3.5-4.5 procs per minute, and abilities, spells and heals 18-23% per cast (the ranges are fits with and without windows whose trigger was not identified). This agrees with community reports of 'about 20%'.

Current implementation / scenario: After the 10-second cooldown, a landed white melee hit grants Clearcasting at 4 procs per minute of attack speed (6.7% in Cat Form, 16.7% in Bear Form), and a landed ability, spell or heal has a 20% chance.

Impact: Feral and Balance resources, and Bear choices where eligible.

Remaining scope: Balance-only samples were small; whether multi-target hits roll once per target.

Implementation / model follow-up: Implemented: after the 10-second cooldown, white hits at 4 procs per minute and other eligible events at 20% (sim/druid/omen_of_clarity.go).

Optional / conditional follow-up

Current beta · level 30
Requires: Level-20+ Druid in Cat Form, then in caster form, with no other Clearcasting source.
Observe / verify: After each proc, count landed white hits and landed abilities or spells separately, from the end of the 10-second cooldown to the next proc. About 60 procs per form.

sim/druid/omen_of_clarity.go · docs/mechanics_review.md · https://wago.tools/db2/SpellAuraOptions?build=1.60.1.69977 · https://wago.tools/db2/Spell?build=1.60.1.69977 · https://foreverlogs.gg/

Earlier references: T25

Link to DRU-001

DRU-004 Thick Hide Defense coefficient Resolved · high

Question: What amount of armor does each Defense point contribute in each form?

Source-backed answer · high confidence: The exact rank Defense-to-Armor coefficient remains sourced. October 8 gear-while-shifted bug notes prompted actual legal weapon-swap controls in Cat/Bear/Moonkin; the active dependency already updates correctly. No duplicate armor buff was added.

Current implementation / scenario: Thick Hide uses the exact sourced rank-dependent Defense coefficient rather than a linear extrapolation. Native rank controls pass.

Impact: Bear gearing, mitigation and normalized defensive-stat value.

Remaining scope: Unmodeled form/gear cases are not certified by the tested legal swaps.

Implementation / model follow-up: Retain existing dependency and source-backed rank controls; current70291 profile/results are refreshed, with publication receipt separate.

sim/druid/forms.go · docs/mechanics_review.md · sim/druid/talents.go · https://wago.tools/db2/TraitDefinitionEffectPoints?build=1.60.1.69977 · https://wago.tools/db2/CurvePoint?build=1.60.1.69977 · https://wago.tools/db2/Spell?build=1.60.1.69977 · https://github.com/ClassicWoWCommunity/forever-bugs/issues/151 · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · tools/forever_bench/druid_hunter_shaman_october9_test.go

Earlier references: Thick Hide follow-up

Link to DRU-004

DRU-013 Bear Swipe's missing server AP term Resolved · high

Question: What is Swipe's intended AP coefficient, and has the announced beta script fix shipped?

Source-backed answer · high confidence: The October 1 official notes give the previously missing numeric coefficient: 3% AP. They explicitly warn that the tooltip remains stale. The supported term is implemented; no remembered SoD coefficient is imported.

Current implementation / scenario: Swipe now adds 3% Attack Power, alongside the existing flat damage and supported modifiers.

Impact: Could change Bear damage and multi-target threat; copying a remembered SoD coefficient would not establish the Forever value.

Remaining scope: No unresolved coefficient question remains. A current-beta AP-only slope check can confirm the deployed server fix but is not needed to guess its intended value.

Implementation / model follow-up: Implemented in sim/druid/swipe.go; refresh affected results separately from source reconciliation.

Optional / conditional follow-up

Sources / code
Requires: A subsequent developer clarification or source-backed updated coefficient.
Observe / verify: Check whether the next patch exposes a numeric AP term; absence from SpellEffect does not disprove a server script.
Current beta · level 30
Requires: Level-16+ Bear after the announced fix ships, known AP deltas, fixed talent multiplier and target armor.
Observe / verify: Measure Swipe damage across AP values, with an SP-only control. Record the build so a still-flat pre-fix result is not mistaken for intended behavior.

https://github.com/ClassicWoWCommunity/forever-bugs/issues/125#issuecomment-5841232704 · https://github.com/ClassicWoWCommunity/forever-bugs/issues/125#issuecomment-5842005903 · sim/druid/swipe.go · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-1/2360696/4 · docs/october-1-corrections.md

Link to DRU-013

DRU-002 Furor's lower-rank Energy carryover Resolved · medium

Question: How are carried Energy, time out of form and the rank-dependent cap combined?

Source-backed answer · high confidence: Furor caps carried Cat Energy by invested rank and preserves elapsed regeneration; it does not add a flat Energy grant on each shift or change Bear form resource rules. Focused carryover/regen/reset checks pass.

Current implementation / scenario: Furor caps carried Cat Energy by invested rank and preserves elapsed regeneration; it does not add a flat Energy grant on each shift or change Bear form resource rules. Focused carryover/regen/reset checks pass.

Impact: Alternate Feral shift builds; current no-Furor profiles do not establish the rule.

Remaining scope: None for the resolved question. Current201-profile results/native-web defaults are refreshed; actual publication is separately evidenced by deploy.yml and live resource checks.

Implementation / model follow-up: Corrected locally with focused native controls; affected exact profiles and published results require refresh before these changes are released.

sim/druid/forms.go · docs/history_review.md · https://wago.tools/db2/Spell?build=1.60.1.69977 · https://wago.tools/db2/TraitDefinitionEffectPoints?build=1.60.1.69977 · https://github.com/ClassicWoWCommunity/forever-bugs/issues/127

Earlier references: T15 (Furor) / H28

Link to DRU-002

DRU-007 Improved Moonfire SP scope Resolved · medium

Question: Does Improved Moonfire multiply only base damage or the SP contribution too?

Source-backed answer · high confidence: Improved Moonfire increases the full direct and periodic damage, including their SP contribution, once; its separate crit bonus remains. Native coefficient controls pass.

Current implementation / scenario: Improved Moonfire increases the full direct and periodic damage, including their SP contribution, once; its separate crit bonus remains. Native coefficient controls pass.

Impact: Balance Moonfire contribution.

Remaining scope: None for the resolved question. Current201-profile results/native-web defaults are refreshed; actual publication is separately evidenced by deploy.yml and live resource checks.

Implementation / model follow-up: Corrected locally with focused native controls; affected exact profiles and published results require refresh before these changes are released.

sim/druid/talents.go · docs/upstream_elliot_review.md · https://wago.tools/db2/SpellEffect?build=1.60.1.70009 · https://wago.tools/db2/TraitDefinitionEffectPoints?build=1.60.1.69977 · docs/check_dispositions.md · sim/druid/moonfire.go

Earlier references: T26 (Improved Moonfire only)

Link to DRU-007

DRU-014 Faerie Fire announced cost/cooldown versus captured data Resolved · medium

Question: Has resource-free six-second Faerie Fire shipped for every form?

Source-backed answer · moderate confidence: Beta players report that 'Faerie Fire does not have a cooldown unless cast in shapeshift form (6s)' (WoW Forever Druid community FAQ (Google Doc linked from the WoW Forever US/NA #druid channel)), and forever-bugs#307 reports the form version's 6-second cooldown wrongly persisting after leaving Cat/Bear Form. Both agree with the captured records: caster Faerie Fire 9907 keeps its mana cost and has no cooldown; the article's free six-second version is the form variant.

Current implementation / scenario: Form variants and captured caster rules remain separate.

Impact: Could change Balance resource use and debuff maintenance.

Remaining scope: None for the simulated cases.

Implementation / model follow-up: Already modelled; no change.

Optional / conditional follow-up

Current beta · level 30
Requires: Caster Faerie Fire access and enough mana for cost observation.
Observe / verify: Record mana delta, cooldown availability and spell ID; isolate Spirit/MP5 ticks.
Beyond level 30 / launch
Requires: Access to later caster and form ranks.
Observe / verify: Confirm the same rules across ranks/forms rather than extrapolating the first rank.

https://news.blizzard.com/en-us/article/24301515/world-of-warcraft-forever-class-deep-dives-hunter-and-druid · sim/druid/faerie_fire.go

Link to DRU-014

DRU-016 Feral bear-weaving and idols after 1.60.1.70425 Resolved · medium

Question: Does bear-weaving with Lacerate gain Feral damage now that Furor keeps Cat Energy through Bear Form?

Source-backed answer · high confidence: Screened on two seeds and two races: weaving variants -0.4% to -33% (shapeshift mana and frozen Energy), and -29% to -36% once Rake is in the rotation; Idol of the Dream +1.2% over Howling Idol.

Current implementation / scenario: No weaving (rechecked with Rake in the rotation). Idol of the Dream (Rip +2 s) is the best idol. Rake is covered by DRU-017.

Impact: Feral.

Remaining scope: None for the current scenario.

Implementation / model follow-up: The Cat spec registers Bear Form and the Bear abilities so weaving APLs can be tested.

sim/druid/druid.go · sim/druid/forms.go

Link to DRU-016

DRU-009 Insect Swarm's final partial tick Outside current scenarios · low

Question: Does the Tier 1 duration extension award a fractional final tick?

Outside current scenarios · moderate confidence: The final partial tick changes the Balance Tier 1 result by less than 0.1%. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Fifteen seconds at a two-second interval yields seven full ticks and one second of remaining aura; no half tick is invented.

Impact: Balance Tier gain.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Beyond level 30 / launch
Requires: The actual Tier 1 duration bonus, not only baseline Insect Swarm.
Observe / verify: Capture the final tick and expiration of a full uninterrupted application; check whether a half-strength tick exists.

docs/forever_tier1.md · sim/druid/insect_swarm.go · https://wago.tools/db2/SpellEffect?build=1.60.1.70009

Earlier references: T11

Link to DRU-009

DRU-015 Furor through Bear Form Resolved · low

Question: Does Furor keep Cat Energy through Bear Form?

Source-backed answer · high confidence: Notes and the new Furor 17056 text; Bear and Dire Bear gain a -100% Energy regeneration aura. Howling Idol's Shifting Power reduction is 0.5 s (1291059). October 16 follow-up: With Furor keeping Cat Energy through Bear Form, bear-weaving was screened (Enrage plus Primal Bite or Lacerate, energy, cooldown and mana gates, with Mangle, Natural Shapeshifter, Enraged Idol and Idol of the Ursine Twins). Every variant lost (the best weaves 0.4-2.8%, frequent weaving 22-33%): each weave pays two shapeshifts (55% of base mana each) that would otherwise fund Shifting Power, and Energy does not regenerate in Bear Form.

Current implementation / scenario: Shifting into Cat restores 20% per point of the Energy last held in Cat plus 2 per point per second spent outside Bear, up to 20 per point; the saved Energy survives Bear Form and Bear time earns nothing.

Impact: Feral (bear-weaving only).

Remaining scope: None.

Implementation / model follow-up: sim/druid/forms.go, sim/druid/items.go.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · https://wago.tools/builds-diff?to=1.60.1.70425&from=1.60.1.70374 · docs/october-15-update.md

Link to DRU-015

Hunter and pets 6 active / 17 total

HUN-001 Pet owner-stat inheritance Needs evidence · high

Question: Do the wiki's AP, SP, hit, crit, Stamina and armor coefficients hold for Hunter and Warlock pets?

Partly answered · moderate confidence: Accept the Forever-specific wiki's measured Hunter inheritance as credible working evidence, rather than leaving every listed coefficient as an unanswered question. The older Classic pages do not rebut it. Full owner-hit transfer and extending all coefficients to Warlock demons remain narrower assumptions because the client scaling auras do not export those values. October 11 audit: Pet tooltip (Hunter Lodge test, 2026-10-06, build 70235): 'Increased by 10% of Master's attack power from whichever is higher: melee or ranged', rounding down, which the implemented inheritance matches. Inheritance is refreshed by a permanent pet aura with a 2-second cooldown (forever-bugs#48 comment). Live bug: with Focused Fire talented, pet stats snapshot at summon and stop updating (forever-bugs#48, 2026-10-11). Unleashed Fury's damage bonus appeared non-functional and Ferocity probably too (same 2026-10-06 test); the engine models the intended talents.

Current implementation / scenario: Explicit Forever wiki-derived inheritance is enabled for both families provisionally. The earlier zero-inheritance notes are superseded. Explicit pet talents are additional, not duplicate inheritance.

Impact: All pet builds; owner gear scaling and pet survivability.

Remaining scope: Warlock-specific power/mana coefficients, full hit-transfer magnitude and dynamic update behavior.

Implementation / model follow-up: No blanket removal of pet inheritance. Separate answered Hunter coefficients from extrapolated demon coefficients in the register.

Answer routes

Current beta · level 30
Requires: Hunter or Warlock pet, one isolated owner-stat change and pet API/damage observations.
Observe / verify: Prioritize demon AP/SP/mana and pet update timing; keep owner and external pet buffs distinct. Do not repeat every accepted Hunter coefficient merely because a server script is hidden.

sim/core/forever_pet_stats.go · docs/check_dispositions.md · docs/upstream_elliot_review.md · https://github.com/classic-hunter/forever-hunter/wiki/Forever-Beta-Changes · https://github.com/ClassicWoWCommunity/forever-bugs/issues/48

Earlier references: T29 (inheritance)

Link to HUN-001

HUN-002 Pet per-hit damage versus speed Needs evidence · high

Question: Do pets of different base speeds deal equal per-hit damage, including AP scaling?

Partly answered · moderate confidence: The Forever-specific wiki and pet-speed issue agree that different speeds currently retain unnormalized per-hit damage. This is adequate evidence for the present equal-hit convention. It does not settle the owner-AP slope at different speeds or establish that the behavior will remain unchanged at launch. October 11 audit: Hunter Lodge 2026-10-06: a 1.2-second cat and a 2.0-second bear at similar pet AP hit for nearly the same minimum damage, so per-hit damage is still not normalized to attack speed, consistent with the equal-hit model.

Current implementation / scenario: Selected Classic-style pet speeds are retained; the equal-hit model uses the adopted reference-speed treatment rather than normalizing DPS across speeds.

Impact: Pet choice and Hunter/Warlock melee contribution.

Remaining scope: The AP-added portion of the hit, especially the fixed two-second reference conversion, and future intended normalization.

Implementation / model follow-up: Retain differing pet speeds and the supported equal-base-hit behavior. Narrow the unknown to AP scaling instead of repeatedly asking whether fast pets exist.

Answer routes

Current beta · level 30
Requires: Two same-family/same-level pets of different speeds and two owner AP totals.
Observe / verify: Compare the change in pet damage per hit across the AP delta for each speed; control happiness and training. This separates an AP coefficient from the already observed base-speed difference.

sim/core/forever_pet_stats.go · sim/hunter/pet.go · docs/check_dispositions.md · https://github.com/classic-hunter/forever-hunter/wiki/Forever-Beta-Changes · https://github.com/ClassicWoWCommunity/forever-bugs/issues/31

Earlier references: T29 (attack speed and damage)

Link to HUN-002

HUN-003 Pet external buffs and action cadence Needs evidence · medium

Question: Which external buffs reach pets, and how quickly does pet AI spend Focus or mana?

Partly answered · high confidence: The no-external-buffs question is answered for both Hunter and Warlock pets by a direct report and an explicit Not A Bug response explaining the removal of pet buff upkeep. Focus base regeneration/cap are already client-backed. Pet AI scheduling is a separate simulation approximation, not missing evidence that pets need buffs. October 11 log audit: Warlock Imp casts in public foreverlogs.gg reports start a median 432 ms after the previous cast ends (see WL-006). Hunter pet abilities were not measured.

Current implementation / scenario: Generic external player buffs are excluded under the adopted common pet model. Focus base regeneration/cap are client-backed; action scheduling is a simulator approximation.

Impact: Pet damage and resource waste; explicit pet talents remain valid exceptions.

Remaining scope: Ability scheduling/resource waste, and any explicitly targeted pet-only exceptions.

Implementation / model follow-up: Close the generic external-buff question. Retain AI cadence as a named model approximation with optional sensitivity checks, not a mandatory repeat of buff eligibility tests.

Answer routes

Current beta · level 30
Requires: Only if pet cadence sensitivity proves material; a trained damage ability and a resource trace.
Observe / verify: Compare ability use when resource-limited versus resource-capped. Do not mix this with owner inheritance or ordinary spell cooldowns.

sim/core/pet.go · sim/hunter/pet.go · docs/check_dispositions.md · https://github.com/ClassicWoWCommunity/forever-bugs/issues/27

Earlier references: T29 (buff/AI portions)

Link to HUN-003

HUN-011 Hunter’s Mark rank-value anomaly Needs evidence · medium

Question: Why does captured rank three give 98 RAP while rank four gives 71?

Partly answered · moderate confidence: Owned Mark casts and owner-specific Expose Prey pass locally. External/foreign marks do not supply ownership. Captured rank-three/rank-four RAP values conflict with expected progression. October 11 audit: Unchanged in 70338: rank 3 (14324) gives 98 RAP and rank 4 (14325) 71; a cost-identical rank-4 record without a skill line (1213268) gives 110.

Current implementation / scenario: Owned learned ranks use the captured values; melee Hunter maintains its own Mark for Expose Prey.

Impact: Affects mark rank choice and ownership; no rank optimization is inferred.

Remaining scope: A current-build tooltip/stat trace or later corrected client record establishing the intended rank values.

Implementation / model follow-up: Keep source versions and implementation coverage separate; no inferred script formula or automatic preset optimization.

Answer routes

Beyond level 30 / launch
Requires: Access to ranks three and four with fixed ranged AP and no competing Mark.
Observe / verify: Compare target-conditioned ranged AP/damage and tooltip values per rank; record the client build.

sim/hunter/hunters_mark.go · https://www.wowhead.com/forever/spell=14325

Link to HUN-011

HUN-012 Savage Strikes article/client version conflict Needs evidence · medium

Question: Is Savage Strikes a melee-damage modifier or family-specific ability crit?

Partly answered · moderate confidence: The official article says melee damage, while captured effect 696821 and trait text/curve specify 2/4% ability crit. Native coverage now includes the complete registered family, not just Raptor/Mongoose.

Current implementation / scenario: Current captured 70009 family-specific crit; full registered family corrected locally.

Impact: Could change melee Hunter damage; stacking both interpretations is unsupported.

Remaining scope: A newer client or controlled trace confirming the announced redesign is live.

Implementation / model follow-up: Keep source versions and implementation coverage separate; no inferred script formula or automatic preset optimization.

Answer routes

Current beta · level 30
Requires: Savage Strikes access and matched talent-on/off attacks at fixed AP, weapon and target.
Observe / verify: Separate hit-size changes from crit-frequency changes, and record build/ability IDs.

https://news.blizzard.com/en-us/article/24301515/world-of-warcraft-forever-class-deep-dives-hunter-and-druid · sim/hunter/talents.go

Link to HUN-012

HUN-016 Fire trap Attack Power scaling Needs evidence · medium

Question: How much Attack Power do Immolation and Explosive Trap gain, and from which Attack Power?

Credible working answer · moderate confidence: The October 15 notes: 'The Hunter's Fire traps now correctly scale with Attack Power.' The client carries no coefficient. Hunter Lodge testers on release day measured about 10% of Attack Power over a whole Immolation Trap, apparently from whichever Attack Power is higher.

Current implementation / scenario: Provisional: the periodic part of each fire trap gains 10% of the higher of melee and ranged Attack Power over its full duration; Explosive Trap's initial hit is unscaled.

Impact: Survival and Pet/Melee (trap users).

Remaining scope: Coefficient, which Attack Power, and Explosive Trap's initial hit.

Implementation / model follow-up: sim/hunter/immolation_trap.go fireTrapAttackPowerBonus.

Answer routes

Current beta · level 30
Requires: Hunter with Immolation Trap and two gear sets differing only in melee or only in ranged Attack Power.
Observe / verify: Total trap damage per set, about 20 traps each.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · sim/hunter/immolation_trap.go

Link to HUN-016

HUN-004 Summon Hawk guardian model Accepted working answer · high

Question: How do Hawk attacks, two-Hawk coexistence and inherited stats actually work?

Credible working answer · moderate confidence: Independent capture of all 222 encounters in public reports 2695/2701 supports two guardians, replacement, retargeting, minimum one-second opening travel and ranged-hasted assault timing. It finds 18 crits in 1,029 landed assaults, contradicting never-crit. These are October 2 observations, not proof of unchanged October 8 server behavior. The tested lifecycle implementation replaces the one-DoT surrogate; numerical rank-60 calibration remains provisional. October 11 log audit: Public foreverlogs.gg reports (1,312 Hawks) confirm two concurrent guardians, at most 18 seconds of attacks, an opening dive about 1.0 s after the cast that never missed, and assault attacks that use the full melee table (4.9% miss, 4.6% dodge, 1.2% parry, 0.8% block). Assault swings were 2.2 s apart in most cases (1.7 s with extra haste). Assault crit was 4.8% overall: 5.1% for a 26-point and 7.6% for a 21-point Beast Mastery Hunter, but individual talent ranks (Ferocity) are not in the logs.

Current implementation / scenario: Two independent 18-second guardians with global oldest replacement, arrival attacks and owner ranged-speed inheritance. Opening uses always-hit, crit-capable damage after at least one second. Assault calibration is 0.35 × rank base, 2.5-second base swings and 1.5% effective crit; UF applies once, Ferocity assault transfer is unresolved.

Impact: Pet/Melee row and Hawk alternatives; major model-dependent build opportunity.

Remaining scope: Base crit versus Ferocity transfer, RAP transfer and higher-rank scaling.

Implementation / model follow-up: Guardian lifecycle, travel, haste, attribution, reset and proc-isolation regressions pass. Quantify 0.30–0.40 damage and 1–3% effective-crit sensitivity; consider Ferocity assault transfer separately. Revalidate Hunter choices rather than treating the calibrated default as measured level-60 data.

Optional / conditional follow-up

Level-30 beta
Requires: Level 25+ Hunter with Summon Hawk; fixed target, RAP and pet settings.
Observe / verify: Summon once, twice and a third time before expiry. Separate opening damage from each guardian's later events and record despawns. Repeat with different owner RAP/crit without changing the target.
Beyond level 30 / launch
Requires: Higher Hawk ranks and the actual role Tier effects.
Observe / verify: Verify rank progression and Tier interactions after the low-rank guardian model is established.

sim/hunter/summon_hawk.go · docs/upstream_elliot_review.md · https://www.wowhead.com/forever/spell=1293241/summon-hawk · https://github.com/classic-hunter/forever-hunter/wiki/Forever-Beta-Changes · artifacts/question_review/class_evidence.json · docs/october-8-review.md · https://foreverlogs.gg/

Earlier references: T14 / H19

Link to HUN-004

HUN-005 Arcane Shot, Serpent Sting and Volley SP Accepted working answer · medium

Question: What SP coefficients replace the missing client coefficients?

Credible working answer · moderate confidence: Hunter Lodge measurements (2026-09-22/23, ranks 1-3): Arcane Shot scales with 10% RAP and not with spell power; Serpent Sting with 15% RAP in total, 3% per tick, read dynamically. With the client zeroing both spell-power coefficients, this replaces the retained Classic values. Traps (Immolation) showed no AP or SP scaling.

Current implementation / scenario: Forever Arcane Shot: rank base + 10% RAP, no spell power. Serpent Sting: rank base + 3% of current RAP per tick (15% total), no spell power, no snapshot.

Impact: SP-heavy Hunter alternatives and AoE; less important for physical-only loadouts.

Remaining scope: Rank-8/9 confirmation of the same RAP coefficients; Volley.

Implementation / model follow-up: Implemented. Paired impact: Marksmanship +2.9%, Beast Mastery +1.4% (RAP scaling outweighs lost SP). Hunter profiles' Flask of Supreme Power and Greater Arcane Elixir now add no damage; consumable selection should be revisited.

Optional / conditional follow-up

Beyond level 30 / launch
Requires: Volley access above level 30, plus independent SP and RAP changes.
Observe / verify: Measure each tick under separate SP/RAP deltas, keeping talents and target resistance fixed.

docs/beta-pass/hunter.md · sim/hunter/arcane_shot.go · sim/hunter/serpent_sting.go · sim/hunter/volley.go · https://github.com/classic-hunter/forever-hunter/wiki/Forever-Beta-Changes

Earlier references: Hunter missing-coefficient notes

Link to HUN-005

HUN-006 Auto Shot and special-shot timing Resolved · medium

Question: Does current Forever retain the Classic half-second Auto Shot wind-up and interactions with Aimed/Sniper/Multi-Shot?

Source-backed answer · moderate confidence: Three independent sources now agree, and the earlier reading of #76 as a sound report missed its resolution. On forever-bugs#76 (2026-09-23) a reply states 'Confirming that you are meant to be able to Auto Shot during those actions' (Aimed Shot, Multi-Shot, Sniper Shot); the same contributor relays upcoming fix schedules elsewhere (#279). The cited forever-hunter wiki says 'Auto Shot can fire while the Hunter is casting another spell with a cast time', and beta players report the same (Hunter Lodge, 2026-10-06 to 10-09). In the client, Aimed Shot drops the Combat interrupt flag (Era 15, Forever 7), carries DO_NOT_RESET_COMBAT_TIMERS and gains SPELL_ATTR4_IGNORE_COMBAT_TIMER.

Current implementation / scenario: Classic half-second Auto Shot wind-up when idle; in Forever, Auto Shot also fires during Aimed Shot, Multi-Shot and Sniper Shot casts as an instant shot that leaves the special's cast and completion intact. Specials do not reset the ranged timer.

Impact: All ranged Hunter comparisons and haste value.

Remaining scope: The wiki also reports melee attacks resetting the ranged timer (intent unclear); not modelled.

Implementation / model follow-up: Implemented: the three shots carry SpellFlagAllowAutoAttacks under Forever; Auto Shot may cast during them with zero wind-up. Paired impact: Marksmanship +12.9%, Beast Mastery +3.6% on top of the RAP change (totals vs the published engine +16.2% / +5.1%).

docs/auto_attack_audit.md · sim/core/attack.go · sim/hunter/aimed_shot.go · https://github.com/classic-hunter/forever-hunter/wiki/Forever-Beta-Changes · https://github.com/ClassicWoWCommunity/forever-bugs/issues/43 · https://github.com/ClassicWoWCommunity/forever-bugs/issues/76

Earlier references: T13

Link to HUN-006

HUN-007 Pet families and new trained abilities Outside current scenarios · medium

Question: Which new family abilities can be added with verified learning paths and effects?

Outside current scenarios · moderate confidence: No benchmark profile uses these pet families or their new abilities. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Claw, Bite, Lightning Breath, Screech and Scorpid Poison are supported, not every new teaching record. Wolf Howl and Bat Sonic Blast are not active benchmark pets.

Impact: Pet selection completeness; not a reason to grant unselected abilities.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Current family, teaching, child-spell, cost and cooldown records.
Observe / verify: Produce one supported-rank matrix per family and implement only joined learning paths. Separate direct values from scripted behavior.
Current beta · level 30
Requires: A family actually tameable by level 20 and its first trained rank.
Observe / verify: Confirm learning, resource cost and damage/cooldown for that family. This does not establish higher-rank availability.

docs/spell_coverage.md · docs/beta-pass/hunter.md · docs/forever-70009.md · sim/hunter/pet_abilities.go · https://github.com/classic-hunter/forever-hunter/wiki/Forever-Beta-Changes · https://github.com/classic-hunter/forever-hunter/issues/23 · https://github.com/classic-hunter/forever-hunter/issues/39 · artifacts/question_review/correctness_followup.json · https://news.blizzard.com/en-us/article/24301515/world-of-warcraft-forever-class-deep-dives-hunter-and-druid

Earlier references: T29 (family abilities)

Link to HUN-007

HUN-008 Dodge-triggered Mongoose Bite window Resolved · medium

Question: Can a dodge activate Defensive State when its aura is initially inactive?

Source-backed answer · high confidence: An always-active dodge listener can open the first Mongoose Bite window; previously the listener was trapped inside its initially inactive window. The first-dodge regression now passes. The un-attacked reference Hunter does not exercise this trigger.

Current implementation / scenario: An always-active dodge listener can open the first Mongoose Bite window; previously the listener was trapped inside its initially inactive window. The first-dodge regression now passes. The un-attacked reference Hunter does not exercise this trigger.

Impact: Solo/reactive Hunter support, not a demonstrated current ranking gain.

Remaining scope: None for the resolved question. Current201-profile results/native-web defaults are refreshed; actual publication is separately evidenced by deploy.yml and live resource checks.

Implementation / model follow-up: Corrected locally with focused native controls; affected exact profiles and published results require refresh before these changes are released.

sim/hunter/mongoose_bite.go · sim/hunter/talents.go · sim/core/aura.go

Link to HUN-008

HUN-010 Careful Aim / Lethal Attacks talent-string mapping Resolved · medium

Question: Do native and web talent strings decode the current tree in the same order?

Source-backed answer · high confidence: A targeted metadata regression reproduced the declaration-order swap used by FillTalentsProto. Hunter proto order now matches the tree; the existing mapping check passes.

Current implementation / scenario: Corrected declaration order with wire IDs unchanged.

Impact: Changes affected Hunter stats and invalidates pre-correction Hunter comparisons.

Remaining scope: None for the resolved question. Current201-profile results/native-web defaults are refreshed; actual publication is separately evidenced by deploy.yml and live resource checks.

Implementation / model follow-up: Keep source versions and implementation coverage separate; no inferred script formula or automatic preset optimization.

proto/hunter.proto · sim/core/character.go · sim/talents_test.go

Link to HUN-010

HUN-013 Tier 1 five-piece Multi-Shot cooldown Resolved · medium

Question: Does the Hunter Tier 1 five-piece reduce Multi-Shot's cooldown as its tooltip says?

Source-backed answer · high confidence: Bonus spell 1301252's cooldown modifier carries class mask 131072, which matches Aimed Shot; Multi-Shot's mask is 4096. Spell modifiers apply by mask, as the project already follows for Efficiency.

Current implementation / scenario: It reduces Aimed Shot's cooldown by one second only.

Impact: Hunter rotations that use Multi-Shot under forced Tier 1.

Remaining scope: Reopen if the mask changes.

Implementation / model follow-up: Multi-Shot removed from the bonus; tier1_test updated.

sim/hunter/item_sets_forever.go

Link to HUN-013

HUN-014 Hunter flask and elixirs without spell-power scaling Resolved · medium

Question: Which flask and elixirs should Forever Hunter profiles use now that their shots no longer scale with spell power?

Source-backed answer · high confidence: Supreme Power and Greater Arcane Elixir add no damage once Arcane Shot and Serpent Sting lose their spell-power coefficients. Every race compared at seed 20261013: Distilled Wisdom over Supreme Power gives Beast Mastery +1.9% (time out of mana 26 s to 3 s) and Marksmanship +2.4%, and is neutral for Survival and Pet/Melee.

Current implementation / scenario: Flask of Distilled Wisdom; no Greater Arcane Elixir.

Impact: All four Hunter rows.

Remaining scope: None.

Implementation / model follow-up: Profile consumables and the default consumable builder updated.

tools/forever_bench/profiles.go

Link to HUN-014

HUN-015 Queued Raptor Strike and the off-hand miss penalty Resolved · medium

Question: Does queueing Raptor Strike still remove the dual-wield off-hand miss penalty?

Source-backed answer · high confidence: Hunter beta testers reported on September 28 that queueing Raptor Strike no longer removes the off-hand penalty (off-hand glancing blows still occurred while it was queued, which the special-attack table cannot produce). The October 1 notes removed the same effect from Heroic Strike. The engine still removed the penalty for the whole queue.

Current implementation / scenario: No. In Forever a queued Raptor Strike leaves off-hand white swings on the dual-wield table; only Raptor Strike itself uses the special-attack table.

Impact: Dual-wielding Survival and Pet/Melee Hunters.

Remaining scope: None.

Implementation / model follow-up: Implemented for Forever (sim/hunter/raptor_strike.go).

sim/hunter/raptor_strike.go · https://us.forums.blizzard.com/en/wow/t/wow-classic-forever-beta-patch-notes/2360696

Link to HUN-015

HUN-009 Demoralizing Screech target AP debuff Outside current scenarios · low

Question: Does the pet apply its sourced AP reduction as well as damage?

Outside current scenarios · moderate confidence: No benchmark profile uses a Screech pet. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Carrion-only Screech now uses active learned ranks and a separate nearby AP debuff. Sourced level-60 AP reduction is 204; damage, Focus, cooldown, expiry and strongest-effect controls pass. Teaching IDs are not cast IDs.

Impact: Tank/solo mitigation with a Screech pet, not selected Cat damage.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Screech pet and controlled target AP with another AP-reduction provider.
Observe / verify: Check rank-level growth, 30-second expiry, area coverage and strongest-effect handling; verify that disabling the selected pet removes only its effect.

sim/hunter/pet_abilities.go · docs/beta-pass/hunter.md · artifacts/question_review/class_evidence.json · https://www.wowhead.com/forever/spell=24579/demoralizing-screech · sim/hunter/demoralizing_screech.go

Link to HUN-009

HUN-017 Hunter changes in 1.60.1.70425 Resolved · low

Question: Which Hunter mechanics changed?

Source-backed answer · high confidence: Client: Multi-Shot 2643 cast time 500 ms to 0 (testers confirm it fires on the move); Mortal Shots 19485 and Efficiency 19416 gain Sniper Shot's class-mask bit. Notes: Expertise, pet crit, Resourcefulness and Trueshot Aura.

Current implementation / scenario: Multi-Shot is instant in Forever. Mortal Shots and Efficiency cover Sniper Shot. Pets inherit the owner's Expertise and take the higher of melee and ranged crit (the sim's single physical crit stat). Resourcefulness procs only from melee and ranged ability crits. Trueshot Aura is raid-wide.

Impact: Marksmanship, Beast Mastery, Survival, Pet/Melee.

Remaining scope: None.

Implementation / model follow-up: sim/hunter.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · https://wago.tools/builds-diff?to=1.60.1.70425&from=1.60.1.70374

Link to HUN-017

Mage 2 active / 11 total

MAG-004 Hot Streak interruption and new-charge ordering Needs evidence · medium

Question: What happens if accelerated Pyroblast is interrupted or new stacks arrive during its cast?

Partly answered · moderate confidence: Completed-Pyro consumption is supported, and the stale 15-second display was corrected before this review. The current discussion again reports all-stack consumption. That does not settle a newly earned in-flight stack or interruption edge case.

Current implementation / scenario: Completed Pyroblast consumes the Heating Up charge/stacks; the engine and current talent display both use the sourced 20-second duration.

Impact: Fire edge cases; not a reopening of the confirmed consumption fix.

Remaining scope: Whether a newly earned in-flight stack is also consumed; interruption matters only for scenarios that interrupt Pyroblast.

Implementation / model follow-up: No further duration/display fix is needed. Preserve completed-Pyro consumption and keep the narrow ordering question separate from the old persistent-buff bug.

Answer routes

Level-30 beta
Requires: Level 25+ Hot Streak/Pyro build and cast/impact/aura logs.
Observe / verify: Start accelerated Pyro while an earlier Fire projectile is in flight; compare aura stacks at impact and completion. Separately cancel Pyro if interruption support is being evaluated.

sim/mage/talents.go · docs/mechanics_review.md · ui/core/talents/trees/mage.json · https://github.com/ClassicWoWCommunity/forever-bugs/issues/92 · artifacts/question_review/class_evidence.json · docs/october-8-review.md

Earlier references: T23 (remaining edge cases)

Link to MAG-004

MAG-011 Frost Mage Cone of Cold on Winter's Chill Model assumption · medium

Question: Should Frost spend Winter's Chill on Cone of Cold, which needs 10-yard range?

Credible working answer · moderate confidence: The October 15 redesign makes Winter's Chill a Cone of Cold multiplier, so the intended rotation weaves Cone of Cold. Standing at 10 yards is a positioning assumption that raid mechanics may not allow.

Current implementation / scenario: Frost stands at 10 yards and casts Cone of Cold at 5 Winter's Chill stacks (+200%); Winter's Chill 4/5 and Arcane Subtlety 1/2.

Impact: Frost (+5.0% confirmed on all races).

Remaining scope: Whether raid encounters allow 10-yard Frost positioning.

Implementation / model follow-up: APL and preset distance 10.

ui/mage/apls/forever_frost.apl.json · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6

Link to MAG-011

MAG-001 Ice Lance coefficient and frozen multiplier Accepted working answer · high

Question: What is the Ice Lance SP coefficient at each rank, and does the frozen multiplier cover the whole hit?

Credible working answer · moderate confidence: Processed public event reports 2727/2687 independently favor about 0.10–0.11 over the old 0.143 guess. Frostbolt-derived SP and repeated low Ice Lance hits constrain the fit. The 0.10 working model replaces that guess, but gear/talent details and resist/absorb fields are incomplete. October 11 audit: WoWSims #forever-mage (2026-10-08/09): Ice Lance rank 2 measured at about 0.113 SP (95% range 0.109–0.118) from 0 versus 58/71 SP logs; rank 1 at about 0.095 (0.089–0.100, 79 hits). Neither matches 1/7. The provisional 0.10 sits between the measured low ranks; the level-60 rank is unmeasured and probabilistic damage rounding widens the intervals. October 15 update: The October 15 notes reduce the frozen bonus to a 200% increase.

Current implementation / scenario: Uses a provisional 0.10 SP coefficient; against frozen targets Ice Lance deals 200% more (x3, was x4 before 1.60.1.70425). Low-rank calibration is extrapolated to the level-60 registration.

Impact: Frost and Arcane–Frost, including comparisons whose baseline casts Ice Lance.

Remaining scope: Controlled SP-delta slope, frozen-state controls and higher-rank extrapolation.

Implementation / model follow-up: Provisional 0.10 applied with a focused registration/SP-delta check. Revalidate affected Mage comparisons; this does not prove an exact level-60 coefficient.

Optional / conditional follow-up

Current beta · level 30
Requires: Level-20 Mage with Ice Lance, two SP totals and a rootable controlled target.
Observe / verify: Compare noncrit damage slopes while unfrozen and frozen. Keep rank, target level and other talents fixed; distinguish a whole-hit multiplier from base-only damage.
Beyond level 30 / launch
Requires: Higher learned ranks if their coefficients differ from the tested rank.
Observe / verify: Repeat only the unresolved rank-scaling part, not the entire frozen test.

sim/mage/ice_lance.go · docs/beta-pass/mage.md · ui/core/talents/trees/mage.json · docs/talent-check-2026-09-17.md · docs/october-8-review.md · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6

Earlier references: T49

Link to MAG-001

MAG-002 Fingers of Frost in-flight timing Resolved · high

Question: Does a cast begun before a chill proc benefit from or consume the resulting charge?

Source-backed answer · low confidence: Public foreverlogs.gg reports show every Frostbolt-triggered Fingers of Frost (121 of 121 outside encounter-start snapshots) at the Frostbolt cast-success timestamp and none at impact. Charges are spent at the next Frostbolt or Ice Lance cast success. Because the proc arrives before the next cast starts, there is no in-flight case for Frostbolt.

Current implementation / scenario: Frostbolt and Frostfire Bolt roll Fingers of Frost when the cast completes and the roll does not miss, before the projectile lands. The cast that grants the charges does not spend them.

Impact: Frost/Arcane–Frost proc usage and apparent hybrid gains.

Remaining scope: None for the simulated rotations.

Implementation / model follow-up: Implemented (sim/mage/talents.go, frostbolt.go, frostfire_bolt.go).

Optional / conditional follow-up

Level-30 beta
Requires: Level 30, one point in Fingers, Frostbolt and precise cast/impact/aura timestamps.
Observe / verify: Compare waiting for impact with immediately starting the next cast, at short and long range. Count charge removal and Shatter outcomes for the already-started cast.
Beyond level 30 / launch
Requires: Level 31 for the second charge, unless later beta access is announced.
Observe / verify: Repeat with two charges; do not assume the one-charge observation establishes simultaneous refresh behavior.

sim/mage/talents.go · docs/beta-pass/mage.md · ui/core/talents/trees/mage.json · https://www.wowhead.com/forever/spell=400647/fingers-of-frost · https://foreverlogs.gg/

Earlier references: T10 / T49 (Fingers timing)

Link to MAG-002

MAG-009 Arcane Blast and Arcane Wrath redesign Accepted working answer · high

Question: How do the new Arcane Blast stacks and Arcane Wrath behave?

Credible working answer · moderate confidence: Notes and client 1.60.1.70425 (Arcane Blast 1239700, Arcane Wrath 1326079-1326082). With these rules the Arcane rotation stacks Arcane Blast first and fills with Arcane Missiles; Arcane-Frost stacks Arcane Blast before Frostbolts.

Current implementation / scenario: Arcane Blast: 1.5 s, 218-254 at 0.429; each cast adds 15% of Intellect as Spell Damage and +175% Arcane Blast cost, up to 4 stacks, 15 s from the first application, not spent by other spells. Arcane Wrath: 3.5 s cast, 6 s cooldown, rank 4 406-496 at 1.0 plus Spell Damage equal to 5% of maximum mana.

Impact: Arcane and Arcane-Frost.

Remaining scope: Whether the stacks' Spell Damage snapshots on Arcane Missiles ticks; Arcane Wrath per-level growth inside a rank.

Implementation / model follow-up: sim/mage/arcane_blast.go; ui/mage/apls/forever_arcane*.apl.json.

Optional / conditional follow-up

Current beta · level 30
Requires: Mage with Arcane Blast and 4 stacks.
Observe / verify: Frostbolt damage with 0 and 4 stacks; stack duration and cost per stack.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · sim/mage/arcane_blast.go

Link to MAG-009

MAG-010 Fire Power and Winter's Chill redesigns Accepted working answer · high

Question: Which effects count for Fire Power, and what does Winter's Chill do now?

Credible working answer · moderate confidence: Client trait ranks and the notes. The notes also carry a second, different Winter's Chill text (a 5% Cone of Cold debuff plus 7% spell damage per stack); client 1.60.1.70425 (aura 271, 40% per stack) is followed.

Current implementation / scenario: Fire Power: 1/2/3% more Fire damage per Fire effect of the mage on the target (Ignite, Improved Scorch, Fireball and Pyroblast DoTs, Impact), up to 4; Frostfire Bolt benefits but does not count. Winter's Chill: +40% to the next Cone of Cold per stack, up to 5, consumed by it.

Impact: Fire (Fire Power) and Frost (Winter's Chill no longer adds crit: about -13% for Frost).

Remaining scope: Which Winter's Chill text is live; whether Improved Scorch counts per target.

Implementation / model follow-up: sim/mage/talents.go.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · sim/mage/talents.go

Link to MAG-010

MAG-003 Frostfire cross-school scripts Outside current scenarios · medium

Question: Which Fire/Frost talents and Tier effects stack on Frostfire beyond explicitly named client masks?

Outside current scenarios · moderate confidence: No benchmark rotation casts Frostfire Bolt in a way that depends on these scripts. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Frostfire direct and periodic talent scopes are separated: Arcane Power/Instability do not raise its DoT; Piercing Ice gives the periodic effect a flat 2%. Existing direct additive stacking and other cross-school rules remain.

Impact: Alternate Frostfire hybrids, not the full-rank Arcane–Frost identity by itself.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Current spell-family masks and the individual role Tier descriptions.
Observe / verify: Enumerate each proposed Frostfire modifier and its positive/negative mask match; isolate genuine script-only gaps.
Beyond level 30 / launch
Requires: Frostfire first learned at level 40 and each disputed talent/Tier effect.
Observe / verify: Test only the remaining script interactions independently at full rank with fixed school power.

sim/mage/frostfire_bolt.go · docs/forever_tier1.md · docs/build_updates.md · sim/mage/talents.go · docs/october-8-review.md

Earlier references: T10

Related: MAG-002 · DATA-001

Link to MAG-003

MAG-005 Clearcasting and Missile Barrage event ordering Accepted working answer · medium

Question: How do channel ticks, last-charge consumption and simultaneous procs interact?

Credible working answer · moderate confidence: Public foreverlogs.gg reports show Missile Barrage gained at the cast-success timestamp of Frostbolt and Fireball (11 of 11) and of Arcane Blast (96), and removed at Arcane Missiles cast success (75 of 96 within 0 ms). October 15 update: Client trait ranks [[20,5,10],[40,10,20]] and the October 15 notes (Missile Barrage moved to row 1 with 2 points; Clearcasting no longer rolls for every target struck).

Current implementation / scenario: Missile Barrage (2 points since 1.60.1.70425) rolls 20/40% from Arcane Blast, 5/10% from each Arcane Missiles hit and 10/20% from Arcane Wrath, Fireball, Frostbolt and Frostfire Bolt, at cast completion. Arcane Concentration rolls once per cast, not once per target struck.

Impact: Arcane and hybrid resources; ordinary persistent-Clearcasting bug is already fixed.

Remaining scope: Clearcasting ordering with Arcane Missiles ticks.

Implementation / model follow-up: Implemented cast-time Missile Barrage for Fireball/Frostbolt/Frostfire Bolt (sim/mage/talents.go).

Optional / conditional follow-up

Current beta · level 30
Requires: Arcane Concentration, Arcane Missiles and mana/aura/tick logging without other proc talents.
Observe / verify: Cast isolated complete and cancelled channels; count Clearcasting procs and identify whether a pre-existing proc is spent at channel start or completion.
Level-30 beta
Requires: Level 25+ Missile Barrage, controlled proc state.
Observe / verify: Compare channels with Clearcasting only, Barrage only and both, including a new proc during the channel. Record actual mana debit rather than tooltip cost.

sim/mage/talents.go · sim/mage/arcane_missiles.go · docs/history_review.md · ui/core/talents/trees/mage.json · docs/october-8-review.md · https://foreverlogs.gg/ · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6

Earlier references: T34

Link to MAG-005

MAG-006 Combustion multi-target charge handling Outside current scenarios · medium

Question: Do multiple hits from one Flamestrike independently add and consume Combustion charges?

Outside current scenarios · high confidence: The open question is per-target charge handling for one area cast. The current Fire reference is single-target and does not use Flamestrike. The implementation TODO does not establish a defect in that reference DPS result.

Current implementation / scenario: The callback handles each landed Fire hit individually and carries an explicit Flamestrike TODO.

Impact: Fire AoE, not the one-target reference rotation.

Remaining scope: Multi-target Fire behavior, if a Fire AoE workload is added.

Implementation / model follow-up: Remove this from the current benchmark-blocking queue while retaining it as an explicit AoE coverage limit.

Optional / conditional follow-up

Beyond level 30 / launch
Requires: Level 40 Combustion and a future multi-target Fire scenario.
Observe / verify: Compare aura charge changes for one Flamestrike hitting one versus several enemies; distinguish noncritical hits, critical hits and periodic ticks.

sim/mage/talents.go · artifacts/forever_input_profiles.json

Link to MAG-006

MAG-007 Master of Elements base-cost refunds and batching Resolved · medium

Question: Do free casts refund mana, and can one simultaneous AoE batch refund repeatedly?

Source-backed answer · high confidence: Primary SpellAuraOptions 29074 specifies ProcCategoryRecovery 9ms, versus Arcane Concentration’s 1000ms. The base-cost refund and batch limit were reproduced: free casts returned zero and three same-batch crits refunded three times. Focused corrections now match the sourced model.

Current implementation / scenario: Forever critical eligible hits refund the rank percentage of base mana cost, including free casts, with the sourced 9ms proc cooldown. Classic behavior is unchanged.

Impact: Mage mana pacing and AoE resource accounting.

Remaining scope: No remaining code/source discrepancy for these two rules; client alignment is not a fresh live-server measurement.

Implementation / model follow-up: Base-cost refund includes free Forever casts; the first qualifying crit starts the 9ms cooldown. Resource amount, school/rank scope and Classic retention have focused regressions.

docs/october-8-review.md · sim/mage/talents.go · tools/forever_bench/mage_october8_test.go · https://github.com/ElliotWood/Forever/pull/599 · https://github.com/ElliotWood/Forever/pull/660

Related: MAG-005

Link to MAG-007

MAG-008 Impact direct-fire source scope Resolved · medium

Question: Does Impact proc from periodic Flamestrike or all generic Fire events?

Source-backed answer · high confidence: Previously absent initialization and Classic-prior3/6/9 UI curve are corrected from active node→definition→Curve83091. Direct/periodic/family/immunity/cross-caster controls pass.

Current implementation / scenario: Forever active Impact uses direct landed owner Mage-family Fire hits/procs,3/7/10%,2 seconds stun; default target immunity and separate opt-in shared proc Stun group 10 DR. Periodic Flamestrike excluded; Classic unchanged.

Impact: Opt-in control/utility, not fabricated extra boss damage.

Remaining scope: No separately captured live tooltip override/deployment claim; DR policy/anchor under CORE-017 and broad foreign item/wand exceptions are not manufactured.

Implementation / model follow-up: Keep the source-linked active tree/generator data coherent and avoid reintroducing the old proportional prior.

Optional / conditional follow-up

Current beta · level 30
Requires: Level 30 Impact allocation, eligible target and Flamestrike/direct Fire control spells.
Observe / verify: Separate direct hits from actual periodic damage and immune targets; use CORE-017 for server DR questions.

https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/mage/talents.go · ui/core/talents/trees/mage.json · tools/forever_bench/mage_impact_october9_test.go · https://wago.tools/db2/CurvePoint?build=1.60.1.70291

Related: CORE-017 · DATA-014

Link to MAG-008

Priest 2 active / 7 total

PRI-001 Shadow Word: Death script value 150 Needs evidence · high

Question: What does the additional dummy/script value do?

Game test needed · low confidence: The raw record identifies 150 specifically as effect type 77 (script effect), separate from direct damage and the 10% backlash dummy. It is not a 150% damage modifier in the client effect system. Neither the current tooltip nor the reviewed public evidence supplies its server script.

Current implementation / scenario: Four learned ranks, sourced damage/cost/cooldown, backlash and Early Demise are implemented. No extra execute multiplier is inferred from 150.

Impact: Shadow damage/execute behavior and possible missing scripted behavior.

Remaining scope: What script 77 does, if anything material beyond the described Death behavior.

Implementation / model follow-up: Continue omitting an invented execute multiplier. This is a concrete script gap, not missing rank/cost/coefficient data.

Answer routes

Beyond level 30 / launch
Requires: Death rank one at level 32; fixed SP, talents and target mitigation.
Observe / verify: Compare damage and backlash above/below execute thresholds, lethal/nonlethal hits and cooldown behavior. Subtract ordinary Early Demise and damage-range effects before attributing anything to 150.

docs/spell_coverage.md · sim/priest/shadow_word_death.go · artifacts/question_review/class_evidence.json

Earlier references: T37 (script)

Link to PRI-001

PRI-003 Inner Focus periodic crit behavior Needs evidence · medium

Question: Does a permitted periodic spell retain Inner Focus's crit bonus after consumption?

Partly answered · moderate confidence: A new public controlled low-crit test reports dynamic Inner Focus crit on existing DoTs while the aura remains active, and loss of the bonus after consumption. This is useful current-beta evidence, but the reporter's preferred snapshot behavior is explicitly an expectation, not a confirmed intended rule. The test is possible now because Inner Focus is a level-20 talent.

Current implementation / scenario: Cost and crit use distinct client masks. Eligible cast-specific behavior is retained; Mind Flay, Death and Starshards do not receive the separate crit bonus merely because they qualify for free cost.

Impact: Priest alternate DoT/channel priorities.

Remaining scope: Intended charged-DoT handling and current-build confirmation; the small reported tick sample is not a precise coefficient estimate.

Implementation / model follow-up: Do not silently turn a reported beta exploit into a permanent uncapped DoT-crit strategy. Keep the client crit eligibility mask separate from cost eligibility.

Answer routes

Current beta · level 30
Requires: Level-20 Priest with Inner Focus and a DoT in its crit mask.
Observe / verify: Compare three arms: no aura, aura activated after the DoT and held, and aura consumed by applying the DoT. Record enough ticks and the exact build; separate live observation from intended design.

sim/priest/talents.go · docs/upstream_elliot_review.md · https://github.com/ClassicWoWCommunity/forever-bugs/issues/137 · ui/core/talents/trees/priest.json

Earlier references: T32 (Inner Focus)

Link to PRI-003

PRI-002 Death backlash and Dark Sacrifice modifiers Outside current scenarios · medium

Question: Which damage, resistance, absorb and survival rules affect these self-costs?

Outside current scenarios · moderate confidence: Self-damage is not counted as DPS. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Death uses its implemented health-backlash interpretation; Dark Sacrifice uses unmodified client amounts and cannot crit. Non-tanking DPS does not model a constrained healer budget.

Impact: Priest resources/survival, not enemy damage from self-hits.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Current beta · level 30
Requires: Level-20 Undead Priest with Dark Sacrifice, fixed maximum health and Spirit, uncapped mana, with/without a controlled absorb.
Observe / verify: Record each health and mana delta, then repeat near mana cap and low health. Separate periodic dummy returns from ordinary Spirit regeneration.
Beyond level 30 / launch
Requires: Death at level 32+, and controlled resistance/absorb/damage-reduction states.
Observe / verify: Measure nonlethal, killing-blow and avoided-hit backlash; include lethal self-cost only when evaluating finite-healing survival.

sim/priest/shadow_word_death.go · docs/beta-pass/priest.md · docs/forever-70009.md · sim/priest/dark_sacrifice.go

Earlier references: T37 (backlash) / Dark Sacrifice modifier caveat

Related: SCEN-002 · PRI-001

Link to PRI-002

PRI-004 Divine Spirit's baseline acquisition Resolved · medium

Question: Can Priests train Divine Spirit after its talent was removed?

Source-backed answer · high confidence: Blizzard's official Deep Dive recap explicitly says Divine Spirit, Blessing of Kings and Improved Mark of the Wild are baseline. This resolves intended Divine Spirit availability without requiring a future trainer screenshot. Current client ranks separately give 17/23/33/40 Spirit at levels 30/40/50/60.

Current implementation / scenario: External 40 Spirit coverage is assumed; surviving class-skill rows are not conclusive trainer evidence. Removed Fortitude improvement talents do not multiply the current sourced buff again.

Impact: Caster/Warlock resources and Spirit-based effects.

Remaining scope: A live trainer omission would be a beta implementation bug, not evidence that a Discipline talent is required.

Implementation / model follow-up: Retain the sourced external 40 Spirit at level 60. Close the general availability question; do not retain obsolete improvement talents or add their multiplier again.

sim/priest/priest.go · docs/beta-pass/priest.md · docs/mechanics_review.md · https://news.blizzard.com/en-us/article/24303313/world-of-warcraft-forever-deep-dive-panel-recap · sim/core/buffs.go · artifacts/question_review/class_evidence.json

Earlier references: T33 (availability portion)

Link to PRI-004

PRI-005 Lesser Heal, Heal and helpful Penance registrations Outside current scenarios · medium

Question: Are the newly adjusted healing ranks actual registered healing spells?

Outside current scenarios · high confidence: Healing spells are outside the DPS and tank scenarios. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: No: Lesser Heal/Heal and helpful Penance remain unregistered; source values are captured. Existing damaging Penance registration is not healing coverage.

Impact: Supported future Priest healing and resource-efficiency work, not present DPS validation.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Current healing child rows and a deliberately supported healing workload.
Observe / verify: Register/test ranks, growth caps, SP/mana/timing and helpful-target branch separately from damaging Penance.
Current beta · level 30
Requires: Level 30 Priest with rank1 Penance and available Lesser Heal/Heal, controlled healing power and missing-health targets.
Observe / verify: Record exact helpful spell IDs, noncrit healing/mana and timing; higher ranks2/4 require later access.

https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/priest/penance.go · tools/forever_bench/ranks_october9_test.go · https://wago.tools/db2/SpellEffect?build=1.60.1.70291

Related: SCEN-007 · DATA-013 · DONE-003

Link to PRI-005

PRI-006 Improved Mind Flay escalation and Dark Inspiration Accepted working answer · medium

Question: Does Improved Mind Flay compound or add per tick, and who gains Dark Inspiration's Spirit?

Credible working answer · moderate confidence: Notes: 'Each instance of damage of Mind Flay deals 10/20% more damage than its previous instance.' Read as compounding.

Current implementation / scenario: Each Mind Flay tick deals (1 + 0.1 x points)^(tick - 1) of the first. Dark Inspiration raises the priest's own Spirit 3% per point for 10 s after Mind Blast or a fully channeled Mind Flay (party members are not simulated).

Impact: Shadow.

Remaining scope: Compounding versus linear; whether the last tick of an interrupted channel counts.

Implementation / model follow-up: sim/priest.

Optional / conditional follow-up

Current beta · level 30
Requires: Shadow Priest with 2/2 Improved Mind Flay.
Observe / verify: Damage of each of the three ticks over 50 channels.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · sim/priest/talents.go

Link to PRI-006

PRI-007 Shadow Mind Flay clipping and Smite Shadow Word: Pain Resolved · low

Question: Do the 70425 priest changes alter the rotations?

Source-backed answer · high confidence: Rotation screens after the October 15 changes.

Current implementation / scenario: Shadow channels Mind Flay fully (Improved Mind Flay makes the last tick the strongest). Smite keeps Shadow Word: Pain up after Holy Fire.

Impact: Shadow +0.4%, Smite +1.3% (confirmed on all races with independent seeds).

Remaining scope: None.

Implementation / model follow-up: Ranked profiles and the Smite APLs.

ui/smite_priest/apls/launch.apl.json · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6

Link to PRI-007

Rogue 4 active / 8 total

ROG-001 Mutilate off-hand flat damage and Cold Blood Needs evidence · medium

Question: Does the off-hand penalty include Mutilate's flat bonus, and does Cold Blood cover both strikes?

Partly answered · high confidence: Official text and child no-miss/dodge/parry flags resolve the parent/child gate and both-hand Cold Blood eligibility. Reproductions also found and fixed a second Seal Fate attempt after a failed first-hand proc. Ordinary builder behavior remains covered. October 15 update: The October 15 notes: 'Cold Blood is now correctly consumed when Mutilate lands a critical strike.' The sim already spent it on Mutilate; it now keeps it when neither strike lands.

Current implementation / scenario: One parent avoidance roll gates both Mutilate strikes and two base combo points. Both strikes keep block/crit and proc eligibility; Cold Blood benefits both and is spent when the Mutilate lands (a fully avoided Mutilate keeps it). Flat bonus placement is unchanged.

Impact: Mutilate damage and cooldown value.

Remaining scope: Off-hand flat-bonus/AP details beyond existing checks and the reported transient beta Cold Blood nonconsumption bug.

Implementation / model follow-up: Corrected gate, child outcomes, Cold Blood ordering and one Seal Fate attempt; do not adopt persistent Cold Blood as intended behavior. Refresh Mutilate/APL evidence on both comparison arms.

Answer routes

Level-30 beta
Requires: Level 30 Mutilate/Cold Blood allocation, two known daggers, fixed AP and target armor.
Observe / verify: Record both child hit/crit outcomes with and without Cold Blood. Swap hand placements and vary AP independently to distinguish flat bonus from weapon contribution.
Beyond level 30 / launch
Requires: Higher Mutilate ranks.
Observe / verify: Verify that the higher flat component follows the low-rank ordering; do not repeat already settled proc timing unless a patch changes it.

sim/rogue/mutilate.go · docs/history_review.md · sim/rogue/talents.go · artifacts/question_review/class_evidence.json · docs/october-8-review.md · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6

Earlier references: T38 / H14

Link to ROG-001

ROG-002 Rogue finisher and poison scaling Needs evidence · medium

Question: Do the retained AP coefficients and poison proc probabilities match Forever?

Partly answered · high confidence: The public developer response gives base AP ratios 0.5% for Instant and 0.45% for Deadly. Instant’s missing direct-hit 0.005 AP term is now implemented for Forever and a matched native AP-delta regression passes; Classic is unchanged. Ordinary damage/crit/resistance multipliers apply once after the term. This is a sourced ratio, not independent level-60 damage measurement.

Current implementation / scenario: The public developer response gives base AP ratios 0.5% for Instant and 0.45% for Deadly. Instant’s missing direct-hit 0.005 AP term is now implemented for Forever and a matched native AP-delta regression passes; Classic is unchanged. Ordinary damage/crit/resistance multipliers apply once after the term. This is a sourced ratio, not independent level-60 damage measurement.

Impact: Rogue build comparisons and poison/physical gear value.

Remaining scope: Deadly’s per-tick/per-stack coefficient placement and the separate finisher/bleed formulas remain unresolved. The Instant correction does not settle those questions.

Implementation / model follow-up: Instant’s source-supported correction is complete locally; replace affected Rogue scenarios before publication. Do not add the Deadly ratio at an inferred placement.

Answer routes

Level-30 beta
Requires: Deadly Poison access and a controlled AP delta.
Observe / verify: Measure one-stack ticks and then five-stack ticks to locate the 0.0045 ratio. Audit finishers in separate AP/CP controls rather than mixing poison proc counts with their damage.

docs/beta-pass/rogue.md · sim/rogue/poisons.go · sim/rogue/rupture.go · sim/rogue/eviscerate.go · https://github.com/ClassicWoWCommunity/forever-bugs/issues/100 · https://github.com/ClassicWoWCommunity/forever-bugs/issues/100#issuecomment-5804701503 · tools/forever_bench/evidence_corrections_test.go

Earlier references: T35 (poison/finisher)

Related: CORE-008

Link to ROG-002

ROG-005 Hack and Slash hand scope and ICD Needs evidence · medium

Question: Is the 200-ms extra-attack ICD real, and how do mixed weapon bonuses apply?

Partly answered · high confidence: The 200-ms ICD is explicitly present in SpellAuraOptions for both parent 13960 and triggered aura 1290312. It is no longer merely an inherited assumption. Dagger/fist and mace child auras also have equipped-weapon masks; those masks do not alone settle every mixed-hand, character-wide application. October 11 log audit: Public foreverlogs.gg reports (210 Sword Specialization extra attacks) show no two extra attacks closer than 824 ms within one encounter, consistent with the sourced cooldown. The logs do not identify the hand, so the crit and armor scope remains open.

Current implementation / scenario: Retains the inherited ICD and current hand-type gates. Talent rank amounts are sourced, event/stacking semantics are not fully measured.

Impact: Combat and mixed-weapon alternatives.

Remaining scope: Mixed-hand crit and armor-ignore scope, not the sourced ICD.

Implementation / model follow-up: Retain the 200-ms gate. Narrow the experiment to mixed weapons and any observed contradiction; do not spend a large sample re-estimating a value already supplied by the current client.

Answer routes

Current beta · level 30
Requires: Level-30 Rogue with 5/5 Hack and Slash (20 Combat points first; 26 points are available) and matched dagger/fist, mace and sword pairs.
Observe / verify: Swap MH/OH placement. Compare each hand's crit and armor-sensitive damage, plus a non-weapon spell control; separate a displayed character-sheet bonus from hand-specific outcomes.

sim/rogue/hack_and_slash.go · docs/beta-pass/rogue.md · artifacts/question_review/class_evidence.json · https://www.wowhead.com/forever/spell=1290312/hack-and-slash · https://foreverlogs.gg/

Link to ROG-005

ROG-004 Deadly Poison V acquisition Needs evidence · low

Question: Is the highest poison available without an unavailable raid book?

Source gap · low confidence: Current spell/item records establish Deadly Poison V's existence and values, not its obtainable recipe. The public item page still carries older damage/charge text and is not a current trainer or loot-source confirmation. Normalized equipment does not settle spell acquisition.

Current implementation / scenario: Retained spell/item records do not independently establish acquisition in the allowed content.

Impact: Conditional highest-rank poison availability; normalized item gear does not make unavailable spells trainable.

Remaining scope: A current trainer, vendor, recipe or loot source for rank five in the allowed content.

Implementation / model follow-up: Retain the availability caveat rather than treating the item database as a learning path. Resolve through source data first; if unavailable at launch, use the highest actually learnable rank.

Answer routes

Sources / code
Requires: Current trainer/vendor/recipe or loot-source export.
Observe / verify: Join the learned recipe to an obtainable source; distinguish the crafted poison item from its recipe.
Beyond level 30 / launch
Requires: Level-60 Rogue and relevant trainer/vendor/content access if no source export resolves it first.
Observe / verify: Confirm acquisition without an unavailable raid book and record the obtainable rank.

docs/check_dispositions.md · sim/rogue/poisons.go · https://www.wowhead.com/forever/item=20844/deadly-poison-v

Earlier references: T56 (availability)

Link to ROG-004

ROG-008 Separate poisons and temporary weapon enchants Resolved · high

Question: Can each hand select its poison independently of a stone or oil?

Known implementation work · moderate confidence: Official article explicitly allows poisons alongside temporary enchants. The old single-enum input cannot represent that legal setup.

Current implementation / scenario: Explicit per-hand Rogue poison options coexist with temporary enchants in native/web controls and all27 current exact profiles. Corrected-engine legal-combination comparisons and profile exports are integrated. Historical requests stay distinct.

Impact: All Rogue profiles, weapon effects and gear selection.

Remaining scope: Native proc coexistence, request round-trip and exact-profile migration are complete. Poison charge depletion remains the separate ROG-003 question; this resolution does not claim that mechanic is verified.

Implementation / model follow-up: Separate inputs, reject stale Forever poison-consume selections rather than silently discard them, preserve Classic handling, and compare current legal combinations.

Optional / conditional follow-up

Sources / code
Requires: Native/web current profiles and isolated weapon fixtures.
Observe / verify: Round-trip both-hand poison+enchant combinations; verify separate stats/procs, equipment restrictions and explicit legacy-input rejection.

docs/october-8-review.md · proto/rogue.proto · sim/rogue/rogue.go · https://worldofwarcraft.blizzard.com/en-us/news/24310968

Link to ROG-008

ROG-007 Blade Flurry copied-damage modifiers Outside current scenarios · medium

Question: Should copied damage receive attacker or target modifiers a second time?

Outside current scenarios · moderate confidence: Blade Flurry matters only against several targets; the benchmark is single-target. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: The inherited secondary-damage path explicitly flags possible double dipping; it is not settled by matching one-target tests.

Impact: Combat two-target damage and target-armor differences.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Level-30 beta
Requires: Level-30 Blade Flurry, two controlled targets and a removable damage modifier.
Observe / verify: Compare primary and copied hits against equal and different armor. Toggle one attacker modifier and one target modifier separately; compare ratios rather than unrelated DPS totals.

sim/rogue/talents.go · sim/core/spell_result.go · artifacts/question_review/class_evidence.json

Link to ROG-007

ROG-003 Poison charge depletion Resolved · low

Question: Can poison charges expire during a supported long encounter?

Source-backed answer · high confidence: Client enchant records give Instant Poison VI (enchant 625) 175 charges and Deadly Poison V (2630) 180 charges, each lasting 30 minutes. Benchmark fights apply poisons 66-85 times in five minutes, so a full application cannot run out within a scenario.

Current implementation / scenario: Finite item charges are not consumed. Current client charge counts are retained as data, not runtime limits.

Impact: Long fights and very fast attack/proc setups.

Remaining scope: None for five-minute scenarios.

Implementation / model follow-up: No change; charges are not depleted in the model.

Optional / conditional follow-up

Sources / code
Requires: Per-hand application counters and current coating limits.
Observe / verify: Record per-iteration maxima for baseline and fastest sensitivity scenarios; add finite charges/reapplication before claiming coverage of arbitrarily long encounters.
Current beta · level 30
Requires: A low-rank finite coating and combat/charge logging.
Observe / verify: Only if consumption semantics remain unclear: distinguish eligible weapon swings, proc attempts and landed poison hits, including resisted applications.

sim/rogue/poisons.go · docs/beta-pass/rogue.md · artifacts/forever_dps_5min.json · artifacts/question_review/class_evidence.json

Earlier references: T56 (charges)

Link to ROG-003

ROG-006 Premeditation expiry and Feint threat Outside current scenarios · low

Question: Are temporary combo points expired and Feint's threat reduction applied?

Outside current scenarios · high confidence: These are known missing implementations, not unknown game rules. Current Rogue damage requests do not use Feint or intentionally hold prepull Premeditation points past their expiry; their damage ranking is not threat-limited. Neither omission presently explains the benchmark order.

Current implementation / scenario: Premeditation grants points without the sourced 20-second expiry; Feint does not apply its actual threat drop.

Impact: Prepull/long-idle Rogue setups and threat-constrained encounters, not ordinary sustained damage alone.

Remaining scope: Temporary-point expiry and Feint's negative threat for future long-idle or threat-constrained setups.

Implementation / model follow-up: Retain both in the implementation backlog, outside the current in-game test queue. A Feint implementation also needs to avoid granting damage-proc eligibility merely because its placeholder uses a melee proc mask.

Optional / conditional follow-up

Sources / code
Requires: Temporary CP ownership/expiry and explicit threat accounting.
Observe / verify: Verify spending before expiry, expiration without erasing unrelated points, target/reset behavior and an actual threat decrease from Feint without damage/proc inflation.

sim/rogue/premeditation.go · sim/rogue/feint.go · docs/beta-pass/rogue.md · artifacts/forever_input_profiles.json

Link to ROG-006

Shaman 2 active / 11 total

SHA-004 Owned buff-totem effects and replacement Implementation gap · medium

Question: Does an APL-cast buff totem actually supply its rank-correct party buff and remove its previous owned buff?

Known implementation work · high confidence: This is a provider-model gap with available rank data. Owned Windfury currently creates a tracking aura rather than the rank-correct functional buff; owned Mana Spring likewise does not supply its recorded regeneration. External raid settings are a separate source and must not be silently replaced or duplicated.

Current implementation / scenario: Some owned totems are tracking auras while raid/party settings supply external buffs. Owned Mana Tide/Healing Stream cancellation is implemented; owned Mana Spring regeneration itself remains incomplete.

Impact: Self-sufficient Shaman/tank-party scenarios and low-rank totem choices; externally supplied ranking buffs must not double count.

Remaining scope: An implementation that tracks owned versus external providers and their expiry/replacement independently.

Implementation / model follow-up: Use shared functional buff implementations with provider ownership, current ranks and exclusive categories. Preserve externally supplied reference support; do not grant both Windfury and Grace merely because one source is an APL totem.

Answer routes

Sources / code
Requires: Owned low/high-rank totems plus independent external providers.
Observe / verify: Check actual stats/resource returns, replacement, expiry and recall. Removing an owned provider must not delete a surviving external provider or leave its own effect permanent.

sim/shaman/water_totems.go · sim/shaman/air_totems.go · docs/upstream_elliot_review.md · docs/beta-pass/core.md · sim/core/buffs.go

Link to SHA-004

SHA-011 Totem of the Storm Maelstrom Weapon chance Needs evidence · low

Question: How often does Lightning Bolt trigger Maelstrom Weapon with Totem of the Storm?

Credible working answer · moderate confidence: Client 1.60.1.70425 spell 1291078 says 'a 50% reduced chance' but gives no spell rate.

Current implementation / scenario: Half the main-hand per-hit Maelstrom Weapon chance (10 per minute) per Lightning Bolt hit.

Impact: Enhancement relic choice.

Remaining scope: The actual chance per Lightning Bolt.

Implementation / model follow-up: Provisional half chance.

Answer routes

Beyond level 30 / launch
Requires: Level-60 Enhancement Shaman with Maelstrom Weapon and Totem of the Storm.
Observe / verify: Maelstrom Weapon stacks gained per Lightning Bolt over about 100 casts.

sim/shaman/talents.go

Link to SHA-011

SHA-001 Maelstrom Weapon proc rate Accepted working answer · high

Question: What determines stack frequency and eligible proc events?

Credible working answer · moderate confidence: A 1-point Enhancement Shaman (26 Enhancement points, 3.6-second mace) in public foreverlogs.gg reports gained a stack on 59% of landed white hits and 52% of landed Stormstrikes (55% overall), about 9-10 procs per minute. Classic's 2 procs per minute per point would give 12% per hit. The client lists no per-rank chance, so the rate is taken as rank-independent; the 5/5 benchmark value (10 per minute) is unchanged.

Current implementation / scenario: 10 procs per minute at every rank in Forever; ranks change only the per-stack cast-time and cost reduction. Only Lightning Bolt spends the stacks.

Impact: Enhancement damage, resources and weapon-speed value.

Remaining scope: Weapon-speed dependence (procs per minute versus a flat chance) and confirmation at higher ranks.

Implementation / model follow-up: Implemented 10 procs per minute at every rank (sim/shaman/talents.go).

Optional / conditional follow-up

Current beta · level 30
Requires: Enhancement Shaman with 1/5 Maelstrom Weapon (25 Enhancement points first), a slow and a fast weapon.
Observe / verify: Melee without casting Lightning Bolt; count stack gains per landed melee hit with each weapon. About 150 gains per weapon.
Beyond level 30 / launch
Requires: Enhancement Shaman with 2/5 or more Maelstrom Weapon (more than 26 talent points).
Observe / verify: Repeat the count to confirm that the proc rate does not change with rank.

sim/shaman/talents.go · docs/beta-pass/shaman.md · docs/check_dispositions.md · https://wago.tools/db2/SpellEffect?build=1.60.1.70009 · https://wago.tools/db2/SpellAuraOptions?build=1.60.1.69977 · ui/core/talents/trees/shaman.json · https://foreverlogs.gg/

Earlier references: T21

Link to SHA-001

SHA-002 Instant Maelstrom Lightning Bolt swing reset Resolved · high

Question: Does a five-stack instant Bolt still reset the next melee swing?

Source-backed answer · moderate confidence: Server semantics settle the model. Lightning Bolt and Chain Lightning carry the Combat interrupt flag (SpellInterrupts 15) and SPELL_ATTR6_DOESNT_RESET_SWING_TIMER_IF_INSTANT (0x02000000), and lack DO_NOT_RESET_COMBAT_TIMERS. TrinityCore's Spell::IsAutoActionResetSpell resets combat timers at cast completion unless the spell is instant with that attribute. So an instant (five-stack) cast keeps the swing timer and any positive cast time resets it, exactly as modelled.

Current implementation / scenario: The recent beta evidence supports instant Lightning Bolt preserving melee swings. Instant LB/CL no longer reset a full swing; positive-time casts still defer melee to cast end plus a full swing, with no melee during casts. Maelstrom only discounts/consumes on Lightning Bolt. Native instant/hardcast controls pass.

Impact: Enhancement APL rankings; all corrected research includes the reset.

Remaining scope: A combat trace would only confirm, not change, the rule.

Implementation / model follow-up: Corrected locally with focused native controls; affected exact profiles and published results require refresh before these changes are released.

Optional / conditional follow-up

Beyond level 30 / launch
Requires: Level-39+ Shaman with five-point Maelstrom and a recorded swing timer.
Observe / verify: Cast five-stack Bolt at multiple points in the swing cycle, then compare ordinary and partial-stack casts. Record start, completion and next auto, not only aggregate DPS.

sim/shaman/lightning_bolt.go · docs/auto_attack_audit.md · ui/core/talents/trees/shaman.json · sim/shaman/chain_lightning.go

Earlier references: T48

Related: SHA-001

Link to SHA-002

SHA-007 Baseline Weapon Mastery grant Resolved · high

Question: What rank/value is granted automatically, and at which levels?

Source-backed answer · moderate confidence: Blizzard retracted it. On ClassicWoWCommunity/forever-bugs#280 (2026-10-08) a reply relays: 'there was a mix up in the article when it was handed off. So it shouldn't have had shield spec and weapon mastery in the baseline portion' (Not A Bug). The 70338 Shaman tree has no Weapon Mastery talent, and players found no damage change on the character sheet. The engine's baseline 10% was based on the article and is removed.

Current implementation / scenario: No baseline Weapon Mastery in Forever; the Forever tree has no Weapon Mastery talent either.

Impact: Enhancement damage and lower-level validation.

Remaining scope: Reopen only if a later build grants it.

Implementation / model follow-up: Removed applyBaselineWeaponMastery; test renamed TestShamanHasNoBaselineWeaponMastery. Enhancement and Stormcaller gear was re-searched without it.

Optional / conditional follow-up

Current beta · level 30
Requires: Shaman trainer/spellbook access, known weapon with no procs and matched target.
Observe / verify: Capture passive grant and compare fixed physical attacks versus magic spells through available levels.
Beyond level 30 / launch
Requires: Level60Shaman with current passives.
Observe / verify: Verify full10%physical weapon bonus and weapon-swap eligibility.

docs/october-8-review.md · sim/shaman/talents.go · https://worldofwarcraft.blizzard.com/en-us/news/24302097

Link to SHA-007

SHA-009 Shaman Flurry duration in swings Resolved · high

Question: How many white swings does a Shaman Flurry proc last?

Source-backed answer · high confidence: The client gives Flurry (16257-16280) three charges, but in public foreverlogs.gg reports all 28 Shamans with Flurry windows lost the buff to the first white swing after its last application or refresh: 1,151 of 1,171 windows ended on that swing, 18 ended without a white swing and 2 contained a second one. Windfury and Stormstrike attacks inside a window did not end it.

Current implementation / scenario: One. In Forever, Flurry ends on the first white swing after the crit that applied or refreshed it. Windfury and Stormstrike attacks do not end it.

Impact: Enhancement attack speed and every Enhancement result.

Remaining scope: Whether this is intended or a server bug that may be fixed; Warrior Flurry is unmeasured (WAR-006).

Implementation / model follow-up: Implemented for Forever: Shaman Flurry has one charge (sim/shaman/talents.go).

sim/shaman/talents.go · https://foreverlogs.gg/

Related: WAR-006

Link to SHA-009

SHA-003 Windfury Weapon extra-attack handling Resolved · medium

Question: Are its proc attacks ordinary extra swings or yellow attacks, with which trigger/cooldown rules?

Source-backed answer · high confidence: Public foreverlogs.gg reports agree with the engine on every point checked: 22.0% procs per landed white swing (19.4% per white swing; 1,283 procs), two extra attacks per proc, a 1.5-second cooldown between procs on one weapon (smallest same-weapon gap 1,493 ms), and full melee outcomes (miss, dodge, parry, block, crit) on the extra attacks.

Current implementation / scenario: Retains the existing proc implementation; no dual-wield Shaman or off-hand attack is granted. Totem is a separate buff and cannot proc on the owner's Windfury-imbued weapon.

Impact: Enhancement weapon/proc value; not the same question as Totem charges.

Remaining scope: Rank/AP routing at level 60.

Implementation / model follow-up: Already modelled; no change.

Optional / conditional follow-up

Level-30 beta
Requires: Level-30 Windfury Weapon, a known two-handed or one-handed MH weapon, no other extra-attack effect.
Observe / verify: Record the proc's two damage events, misses/crits/glances, regular swing timing and proc spacing. Change AP independently and keep the initial rank distinct from its max-scaled tooltip.
Beyond level 30 / launch
Requires: Later imbue ranks and the actual Windfury Totem provider.
Observe / verify: Confirm rank/AP routing and that an own Windfury weapon excludes the separate Totem benefit.

sim/shaman/windfury_weapon.go · docs/beta-pass/shaman.md · https://www.wowhead.com/forever/spell=439431/windfury-weapon · artifacts/question_review/class_evidence.json · https://foreverlogs.gg/

Earlier references: T30 (Weapon)

Link to SHA-003

SHA-006 Flametongue Totem damage formula and comparison Outside current scenarios · medium

Question: What numeric on-hit rule is needed to compare Flametongue + Grace with Windfury?

Outside current scenarios · moderate confidence: The default profiles do not use Flametongue Totem. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Opt-in zero-SP, provisional rank dummy/100 × base MH-speed proc. Grace and Windfury Weapon may coexist; Windfury Weapon child extra attacks no longer also trigger it. Windfury Totem/Flametongue Weapon remain conflicting. Canonical Windfury support is unchanged.

Impact: May change melee raid support. A provisional comparison is possible, but does not verify the server formula.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: A current script explanation or controlled Forever source with a numeric formula.
Observe / verify: Resolve weapon-speed damage, SP contribution and eligible hits without treating a dummy field as damage.
Level-30 beta
Requires: Rank-one Flametongue at level 28; several weapon speeds and known SP deltas.
Observe / verify: Measure on-hit fire damage and eligible white/special hits, removing other damage procs. Grace of Air is not available in this stage.
Beyond level 30 / launch
Requires: Full-rank Flametongue, Grace of Air and Windfury; one fixed melee profile.
Observe / verify: Compare exclusive FT+Grace versus Windfury with the sourced model and matched inputs; rank-one evidence is not full-rank validation.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-updated-september-24/2360696 · https://www.wowhead.com/forever/spell=16387 · https://warcraft.wiki.gg/wiki/Flametongue_Totem · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-1/2360696/4 · docs/october-1-corrections.md · docs/october-8-review.md · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · tools/forever_bench/druid_hunter_shaman_october9_test.go

Link to SHA-006

SHA-010 Elemental changes in 1.60.1.70425 Accepted working answer · medium

Question: How are the new Lava Burst ranks, Elemental Mastery and Call of Flame modelled?

Credible working answer · moderate confidence: Client spell data and the notes.

Current implementation / scenario: Lava Burst ranks at 30/40/50/60 (rank 4, 1326047: 192-248, 265 mana). Elemental Mastery: 15 s, Lightning Bolt and Lava Burst 0.5 s faster, no Chain Lightning cooldown, 3 min cooldown. Call of Flame adds 3/6/9 s to Flame Shock. Chain Lightning and Fire Nova cost less; Fire Nova 0.143. Stormstrike's charge is spent only by a spell that lands.

Impact: Elemental, Stormcaller, Enhancement.

Remaining scope: Lava Burst growth per level inside a rank.

Implementation / model follow-up: sim/shaman.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · https://wago.tools/builds-diff?to=1.60.1.70425&from=1.60.1.70374

Link to SHA-010

SHA-005 Fire Nova threat owner Outside current scenarios · low

Question: Is Fire Nova threat attributed to the Shaman or the active Fire totem?

Outside current scenarios · high confidence: The unresolved issue is threat ownership, not Nova damage or its owned-Fire-totem requirement. No current Shaman tank row is ranked, and stationary Shaman DPS has no aggro constraint. Changing ownership would not alter that damage comparison.

Current implementation / scenario: Requires a live owned Fire totem; damage and threat are credited to the Shaman.

Impact: Threat-constrained encounters, not damage amount in a stationary unthreatened DPS run.

Remaining scope: Threat attribution for a future Shaman tank or aggro-constrained workload.

Implementation / model follow-up: Keep the limitation visible but remove it from the current DPS-blocking queue. Do not change the verified live-totem requirement.

Optional / conditional follow-up

Current beta · level 30
Requires: Fire Nova and a live Fire totem, an isolated enemy and a threat API observer.
Observe / verify: When threat-constrained support is needed, measure owner and totem threat before/after Nova with no other damage; repeat after totem replacement.

sim/shaman/fire_totems.go · docs/mechanics_review.md · artifacts/forever_input_profiles.json

Earlier references: T27

Link to SHA-005

SHA-008 Elemental Devastation trigger events Resolved · low

Question: Which spell critical strikes trigger Elemental Devastation?

Source-backed answer · high confidence: Blizzard (forever-bugs#257, 2026-10-06): 'This activates off of Harmful Spell Critical. Not periodic, and not critical strikes from procs (flametongue), or proxies (totems).' The engine already excluded periodic damage, passive procs and totem attacks (empty proc mask); Fire Nova, which Forever casts from its totem, still triggered it.

Current implementation / scenario: Direct harmful Shaman spell crits only: not periodic ticks, weapon-imbue procs, totem attacks or Fire Nova.

Impact: Enhancement and Stormcaller melee crit; no current benchmark rotation casts Fire Nova.

Remaining scope: None.

Implementation / model follow-up: Fire Nova excluded under Forever.

sim/shaman/talents.go · https://github.com/ClassicWoWCommunity/forever-bugs/issues/257

Link to SHA-008

Warlock 4 active / 12 total

WL-001 Demonic Brand power source and threat Needs evidence · high

Question: Whose school power does Brand read, can it crit, and what threat multiplier applies?

Partly answered · moderate confidence: October 8 corroborates already-published Fire-normal/Shadow-high threat separation. No additional Fire threat multiplier is applied in this update. October 11 log audit: Community testers derived the damage formula from the client files as (level - 26) x 1.5 + a 14-17 roll + 7.8% of the Warlock's spell power, which is the formula modelled, and Brand hits can crit. Public foreverlogs.gg reports contain only 11 Brand hits, too few to measure.

Current implementation / scenario: Owner-school power feeds ranked target-scoped Brand charges, with pet modifiers once. Both children always hit and can crit. Fire uses normal threat; Shadow retains the existing 3× high-threat model.

Impact: Demonic Pact pet damage and threat; its formula is not just flat demo damage anymore.

Remaining scope: Shadow absolute high-threat convention and rank/power/charge behavior beyond sourced/checked scopes remain qualified.

Implementation / model follow-up: Retain the previous correction and direct source attributes. Current full201-profile benchmark is refreshed; deployment status is a separate workflow/live receipt.

Answer routes

Level-30 beta
Requires: Level 25+ Brand, Imp and a melee demon, controlled owner school power and pet modifiers.
Observe / verify: Separate normal pet attacks from Brand events. Vary owner power, record Brand crits and threat, and test a second target without a Brand.
Beyond level 30 / launch
Requires: Max-level pet ranks and relevant Tier effects.
Observe / verify: Verify the level extrapolation and Tier interactions rather than rediscovering the basic power owner.

sim/warlock/talents.go · docs/history_review.md · https://www.wowhead.com/forever/spell=1293697/demonic-brand · artifacts/question_review/class_evidence.json · docs/october-8-review.md · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md

Earlier references: T39 / H24

Link to WL-001

WL-009 Demonic Knowledge ranked values Needs evidence · high

Question: Does up to 33% in the article mean the full talent, or rank-one shorthand?

Partly answered · moderate confidence: The new official narrative says up to 33%; unchanged trait curve 83192 has 33/67/100. No current controlled rank comparison settles the conflict.

Current implementation / scenario: Preserves the static 33/67/100% oflevel curve for owner and active demon.

Impact: Demonology talent value and pet spell damage.

Remaining scope: Observed spellpower delta at each talent rank and on owner/pet independently.

Implementation / model follow-up: Do not silently divide the full talent by three or label the unchanged curve as newly datamined.

Answer routes

Current beta · level 30
Requires: Level-30 Warlock with 1/3, 2/3 and 3/3 Demonic Knowledge (20 Demonology points first; 26 points are available), fixed owner gear and an active pet.
Observe / verify: Measure owner and pet spell power at each rank, despawning and resummoning after each change.
Beyond level 30 / launch
Requires: Higher-level Warlock.
Observe / verify: Confirm how the bonus scales with player level.

docs/october-8-review.md · sim/warlock/talents.go · https://worldofwarcraft.blizzard.com/en-us/news/24310968

Link to WL-009

WL-002 Incinerate's Immolate bonus scope Needs evidence · medium

Question: Does the 25% Immolate bonus multiply base damage only or the whole SP-scaled hit?

Game test needed · low confidence: The current client encodes the Immolate bonus as a 25-valued dummy, not a standard final-damage multiplier. The engine multiplies only the random base before adding SP. The public tooltip repeats the bonus but does not resolve that ordering; no controlled Forever result was found in the reviewed public sources.

Current implementation / scenario: Retains the current implementation pending a controlled coefficient test; tooltip wording alone was insufficient.

Impact: Destruction and alternate Fire rotations.

Remaining scope: Base-only versus full-hit multiplier.

Implementation / model follow-up: Do not import another expansion's formula. A two-SP-level comparison can settle this directly.

Answer routes

Beyond level 30 / launch
Requires: Level-40 Incinerate and two known SP totals on the same target.
Observe / verify: Measure with/without Immolate at each SP total. Compare the slopes, not just one tooltip-range average; retain identical damage talents and mitigation.

sim/warlock/incinerate.go · docs/check_dispositions.md · https://www.wowhead.com/forever/spell=412758/incinerate · artifacts/question_review/class_evidence.json

Earlier references: T08

Link to WL-002

WL-006 Improved Imp's extra dummy effect Needs evidence · medium

Question: Does the undocumented -300/-700/-1000 effect shorten Firebolt casts?

Partly answered · moderate confidence: The active trait does contain the -300/-700/-1000 dummy, but neither current prose nor effect type identifies its script. It cannot be safely converted into cast-time milliseconds. Separately, the documented Firebolt bonus is a normal damage modifier, while the code scales only base damage; that distinct implementation issue is recorded as WL-008. October 11 log audit: Public foreverlogs.gg reports (2,443 Imp cast cycles) show a 2.0-second Firebolt cast followed by a median 432 ms pause (10th-90th percentile 389-467 ms) before the next cast starts. The engine's artificial pause was 200 ms; Forever now uses 432 ms. The talent ranks of those Imps are unknown, so the Improved Imp dummy effect is still unexplained; the observed casts were not shortened.

Current implementation / scenario: The suspected cast-time modifier is not inferred from the dummy values alone.

Impact: Live Imp alternatives; sacrificed Imps do not gain combat damage.

Remaining scope: Whether Improved Imp changes the cast or the pause (compare 0/3 and 3/3 with the same gear).

Implementation / model follow-up: Implemented the measured 432 ms pause for Forever (sim/warlock/imp.go).

Answer routes

Current beta · level 30
Requires: A live Imp, ranks 0/1/2/3 of Improved Imp, fixed owner/pet haste and mana.
Observe / verify: Record Firebolt cast start/end and the interval before the next cast. Distinguish a shorter cast from removal of the engine's artificial 200-ms recast delay.

sim/warlock/imp.go · docs/beta-pass/warlock.md · https://www.wowhead.com/forever/spell=18694/improved-imp · artifacts/question_review/class_evidence.json · https://foreverlogs.gg/

Related: WL-008

Link to WL-006

WL-010 Bane of Havoc damage copying Resolved · high

Question: Are copy scope, one-target limit and threat represented without duplicated modifiers?

Source-backed answer · moderate confidence: A tester's public Warcraft Logs analysis (Warlock Classic Discord, 2026-10-05) matches the implemented model point for point: Bane of Havoc copies 15% of damage dealt to other targets (including AoE) to the Havoc target; pet damage and Hellfire self-damage do not trigger it; the copy cannot crit but copies 15% of a crit; it does not double-dip Curse of the Elements; overkill counts in full; wand Shoot triggers it.

Current implementation / scenario: Copies 15% of actual qualifying owner damage as a separate Shadow event to the marked enemy, with a per-owner one-target limit. No recursive copies, pet/self/healing leakage or repeated damage modifiers.

Impact: Multi-target Warlock damage; no gain in the current one-target reference.

Remaining scope: None for the simulated cases.

Implementation / model follow-up: Already implemented; no change.

Optional / conditional follow-up

Sources / code
Requires: Two targets and two Warlocks with independent Havoc/Banes.
Observe / verify: Check15%copy, expiry/replacement/caster ownership, direct and periodic damage, no duplicated mitigation/crit and no recursive proc chains.
Current beta · level 30
Requires: Havoc is available on the test character; otherwise defer access.
Observe / verify: Measure copy and threat behavior with separate owner/pet/self-damage events and mitigation changes.

docs/october-8-review.md · sim/warlock/curses.go · https://worldofwarcraft.blizzard.com/en-us/news/24310968

Link to WL-010

WL-003 Wrack's amplified DoT scope Resolved · medium

Question: Which other Shadow DoTs receive its amplification, and exactly when does cancellation remove it?

Source-backed answer · high confidence: The new official article says Wrack amplifies other owned Shadow DoTs. Unchanged static effect 1349557 family mask 0 x 402 identifies only Corruption/Agony, matching current code and explicit tests excluding Doom/SiphonLife/DrainSoul. Narrative intent and client/script scope genuinely conflict.

Current implementation / scenario: Wrack amplification is confined to the caster’s sourced Corruption/Curse of Agony family; unrelated or another owner’s DoTs do not gain the modifier. Native ownership, family and cancellation controls pass; learn-level metadata is 40.

Impact: Affliction channel/DoT interactions.

Remaining scope: Controlled testing of each additional owned Shadow periodic family; article wording alone does not establish a changed live mask.

Implementation / model follow-up: Keep 10%, caster ownership and cancellation rules; do not silently broaden scope. Test Wrack-talented APLs separately from current no-Wrack Affliction presets after resolving the relevant family model.

sim/warlock/wrack.go · docs/check_dispositions.md · https://www.wowhead.com/forever/spell=1316697/wrack · artifacts/question_review/class_evidence.json · docs/october-8-review.md

Earlier references: T07

Link to WL-003

WL-008 Improved Imp's Firebolt damage multiplier Resolved · medium

Question: Does Improved Imp multiply Firebolt's spell-power contribution as well as its base damage?

Source-backed answer · high confidence: Improved Imp multiplies full Firebolt damage, including its SP component, at all three talent ranks. The separate unexplained dummy effect is still tracked under WL-006.

Current implementation / scenario: Improved Imp multiplies full Firebolt damage, including its SP component, at all three talent ranks. The separate unexplained dummy effect is still tracked under WL-006.

Impact: Undervalues the talent for an active Imp with spell power; this is separate from generic pet inheritance and the undocumented dummy effect.

Remaining scope: None for the resolved question. Current201-profile results/native-web defaults are refreshed; actual publication is separately evidenced by deploy.yml and live resource checks.

Implementation / model follow-up: Corrected locally with focused native controls; affected exact profiles and published results require refresh before these changes are released.

sim/warlock/imp.go · https://www.wowhead.com/forever/spell=18694/improved-imp · artifacts/question_review/class_evidence.json

Related: WL-006

Link to WL-008

WL-011 Siphon Life periodic critical strikes Resolved · medium

Question: Can Siphon Life ticks crit under Forever's periodic-crit rule?

Source-backed answer · high confidence: Forever marks each crit-capable DoT with SPELL_ATTR8_PERIODIC_CAN_CRIT (added to 121 simulated spells). Siphon Life 18265/18879/18880/18881 lacks it, and a player logged 219 ticks with no crit at about 5.6% spell crit (forever-bugs#303). Pandemic's tooltip still lists Siphon Life; the live behaviour follows the attribute. No hotfix in the 70338 DBCache touches it.

Current implementation / scenario: Siphon Life ticks cannot crit; other Warlock DoTs can.

Impact: Affliction and any build that maintains Siphon Life (Affliction −1.2%).

Remaining scope: Reopen if a later build adds the attribute.

Implementation / model follow-up: SpellFlagNoPeriodicCrit on Siphon Life.

sim/warlock/siphon_life.go · https://github.com/ClassicWoWCommunity/forever-bugs/issues/303

Link to WL-011

WL-012 Decimation trigger timing Resolved · medium

Question: Does Shadow Bolt grant Decimation when the cast completes or when the bolt lands?

Source-backed answer · high confidence: Warlock beta testers' combat-log traces show Decimation applied at Shadow Bolt cast completion, provided the bolt hits, not when the projectile lands. The engine previously waited for the projectile.

Current implementation / scenario: In the execute phase, Shadow Bolt grants Decimation when its cast completes, if the bolt does not miss.

Impact: Demonology and Destruction execute-phase Soul Fire timing.

Remaining scope: None.

Implementation / model follow-up: Implemented for Forever: Shadow Bolt applies Decimation at cast completion.

sim/warlock/talents.go · sim/warlock/shadowbolt.go

Link to WL-012

WL-004 Spellstone's Fire/Shadow mask Accepted working answer · low

Question: Does Spellstone increase both Fire and Shadow power despite Shadow-only wording?

Credible working answer · moderate confidence: The official Rogue/Warlock article explicitly allows Warlock stones with Wizard Oil. Native additive coexistence now passes for both stone types and the web conflict warning is removed. This resolves stacking, not every damage-school flag.

Current implementation / scenario: Spellstone grants its modeled Fire/Shadow bonuses and casting speed independently of a legal weapon oil; Firestone similarly stacks. The prior native rejection/UI warning was incorrect.

Impact: Optional weapon imbue, not a current baseline modifier.

Remaining scope: Only a contrary server observation or later corrected effect would require reopening the selected mask.

Implementation / model follow-up: Retain stone effects and compare legal stone+oil choices for current defaults; no blanket spell-power conversion or duplicate bonus.

Optional / conditional follow-up

Beyond level 30 / launch
Requires: Level-36+ Warlock with a Spellstone.
Observe / verify: Fire and Shadow spell damage with and without the Spellstone; about 40 casts per school.

docs/history_review.md · sim/warlock/talents.go · artifacts/question_review/class_evidence.json · docs/october-8-review.md

Earlier references: T42

Link to WL-004

WL-005 Fel Armor acquisition Outside current scenarios · low

Question: Is Fel Armor actually available rather than a retained rune record?

Outside current scenarios · moderate confidence: Fel Armor is not available. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Not granted; the misleading duplicate Demon Armor selector was removed.

Impact: Optional Warlock defensive/resource setup.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: A current trainer/quest/skill-path source identifying this spell, not just a same-name retained row.
Observe / verify: Reopen only when that acquisition chain exists; then verify effects and exclusivity before adding a selectable option.

docs/history_review.md · docs/spell_coverage.md · sim/warlock/warlock.go · https://www.warcrafttavern.com/forever/guides/warlock/

Earlier references: T43

Link to WL-005

WL-007 Soul Shards and kill-triggered resource effects Outside current scenarios · low

Question: Can the simulator represent shard costs/refunds and qualifying kills?

Outside current scenarios · high confidence: Benchmark fights assume unlimited Soul Shards and no kill-triggered effects. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: No finite Soul Shard inventory or full kill-event model; some shard-related talents have no meaningful cost effect.

Impact: Short-fight openers, repeated summons and execute/kill scenarios.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Declared starting shards and sourced per-spell reagent/refund rules.
Observe / verify: Check inventory conservation, cast eligibility at zero, reset and exact qualifying kill events; keep assumed unlimited-shard reference results clearly labeled.

sim/warlock/warlock.go · docs/beta-pass/warlock.md · https://www.warcrafttavern.com/forever/guides/warlock/

Link to WL-007

Equipment, enchants and data coverage 4 active / 18 total

DATA-001 Server downranking penalties Needs evidence · high

Question: Does the server apply additional low-rank SP penalties beyond client coefficients?

Partly answered · moderate confidence: October 8 adds explicit reduced Improved Wing Clip proc chance for substantially downranked casts, without publishing thresholds/formula. Early caster rank smoothing is a base/growth update, not proof that hidden downrank penalties disappeared.

Current implementation / scenario: Captured rank coefficients and new early-rank ranges remain. Active caster presets use highest registered ranks; Hunter Wing Clip likewise registers only its highest trained rank. Additional hidden SP/proc penalties are not invented.

Impact: Current rank-2 Smite/Lightning Bolt, lower-rank Paladin choices and conditional Frostfire research.

Remaining scope: Server penalty curves, including Wing Clip proc reduction and imported low-rank rotations.

Implementation / model follow-up: Disclose limited highest-rank Hunter coverage and preserve separate rank/source/acquisition checks.

Answer routes

Current beta · level 30
Requires: Two learned ranks of a caster damage spell, controlled SP-only gear changes, fixed target level/resistance and no changing damage talents.
Observe / verify: Compare the damage slope per SP at each rank, not their raw hit sizes. Record exact player level and exclude crits; repeat after leveling if possible.
Level-30 beta
Requires: The same spell/rank and controls at level 30.
Observe / verify: Check whether its SP slope changes as the player outlevels the rank. This extends the low-level test, not the level-60 conclusion.
Beyond level 30 / launch
Requires: Level 60 and the actual downranked spells used by the retained APLs.
Observe / verify: Repeat the two-SP slope comparison for low and maximum ranks; separate learned-level base growth from the coefficient.
Current beta · level 30
Requires: Level 30 Hunter with Improved Wing Clip and independently accessible lower/higher rank casts, fixed target eligibility.
Observe / verify: Estimate landed-hit proc probability by rank/player level, holding talent and target fixed; current highest-rank-only registration cannot answer this.

docs/build_updates.md · docs/check_dispositions.md · https://www.60.tools/downrank-calculator · sim/mage/frostfire_bolt.go · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-1/2360696/4 · docs/october-1-corrections.md · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/hunter/wing_clip.go · tools/forever_bench/ranks_october9_test.go

Earlier references: T12

Link to DATA-001

DATA-002 Projected gear budget and stat prices Model assumption · high

Question: How representative is equal modeled item cost across physical, caster and tank stats?

Scenario decision · high confidence: The gear-generation proof reproduces the captured armor/weapon quantity tables, but the allocation exponent, stat prices, Stamina shares and passive-trinket budget are explicit modeling choices. Equal modeled cost is not evidence of equal real-world gearing opportunity or equal DPS. Current requests contain projected equipment, not hidden real-item procs or a paid-hit conversion.

Current implementation / scenario: Hypothetical level-65 items use sourced quantity/armor/weapon tables but assumed nonlinear allocation budgets and special-stat prices. Equal cost is not equal DPS, and Intellect is not generic SP.

Impact: All cross-class rankings and gear scaling; tank defensive coverage must not be fabricated.

Remaining scope: Representativeness of the chosen budget and templates, not an unknown combat mechanic.

Implementation / model follow-up: Keep the spreadsheet, per-item allocation proofs and legal armor constraints visible. Assess alternative prices as separate scenarios; replace assumptions only when enough current item observations constrain them.

Answer routes

Sources / code
Requires: A proposed alternative budget or additional sourced crafted/PvP items.
Observe / verify: Fit out-of-sample items and compare total selected stat budgets by role; do not judge equivalence from the final DPS ranking.

docs/modelled_gear.md · assets/db_inputs/forever_synthetic_gear_v2.json · artifacts/modelled_gear/proposed_gear.xlsx · artifacts/modelled_gear/gear_equity.json

Earlier references: Projected equipment limitations

Link to DATA-002

DATA-018 Relics: real effects with a quarter of the modeled relic stats Model assumption · high

Question: Which relic should the modeled benchmark equip?

Credible working answer · moderate confidence: Relic effects matter more than the small stat budget a level-65 relic would carry, and the real relics are known. A quarter of the modeled slot budget keeps some stats in the slot without inventing a full item.

Current implementation / scenario: Every real Forever idol, libram and totem whose effect the sim implements is available as a projected copy carrying its own effect plus a quarter of the modeled level-65 relic's stats (physical: 2 Stamina, 1 Agility, 6.5 Attack Power; spell: 2 Stamina, 1.75 Intellect, 2.25 Spell Power). The search picks the best one per profile; the full-capacity modeled relics are no longer ranking gear.

Impact: Druid, Paladin and Shaman rows (the relic slot).

Remaining scope: Actual stats on level-65 relics at launch.

Implementation / model follow-up: effect_item_id on items lets a projected copy run the real item's effect.

tools/database/generate_forever_hybrid_relics.py · assets/db_inputs/forever_hybrid_relics.json

Link to DATA-018

DATA-006 Legacy item and enchant proc assumptions Needs evidence · medium

Question: Which inherited handlers still use assumed PPM, triggers or obsolete effect values?

Partly answered · high confidence: Source payload and probability are separate. Recovery's dodge/parry100% event route, five-percent heal and10s gate are established; Revelation payload/form behavior is sourced but its crit-dependent proc script is unpublished. Old Crusader-heal code-ready wording is superseded.

Current implementation / scenario: Crusader/Lifestealing payload corrections are already present; legacy proc frequencies remain qualified. New Recovery is source-backed; AP kits grant once. Revelation needs an explicit profile-owned model and is integrated locally under explicit assumptions, not hidden 5%.

Impact: Manual catalog items and proc enchants on the projected gear pool. Form-specific enchant eligibility is tracked separately.

Remaining scope: Effective rates/eligibility for other legacy effects; Revelation law under DATA-015. Dormant real-item catalog assumptions are not certified.

Implementation / model follow-up: Use actual source chains, reject illegal enchant layouts and keep recipe versus wearer constraints separate. Do not interpret a triggered wrapper100% as per-swing chance.

Answer routes

Sources / code
Requires: Crusader enchant 1900 → spell 20007; fixed proc events at levels 20 and 60.
Observe / verify: Check 100 Strength, the sourced self-heal range, health capping and damage/threat separation. Verify Lifestealing's heal separately if that enchant is enabled.
Beyond level 30 / launch
Requires: Actual Crusader/Fiery enchants, fixed-speed weapons and separate white-only, special-attack and extra-swing samples. Move earlier only if the beta supplies these high-skill enchants.
Observe / verify: Estimate probability by eligible event type and weapon speed; record haste, handedness and form. A triggered spell's coefficient does not establish its PPM.

sim/common/item_effects.go · sim/common/enchant_effects.go · docs/mythicsim_review.md · ui/core/spells/common.json · artifacts/question_review/scenario_data_evidence.json · https://www.wowhead.com/forever/spell=20007/holy-strength · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · tools/forever_bench/enchant_october9_test.go · assets/db_inputs/forever_enchants_70291.json

Earlier references: M08 / legacy proc TODO inventory

Related: DRU-011 · SCEN-002 · DATA-015

Link to DATA-006

DATA-003 Unresolved real-item stats and acquisition Outside current scenarios · medium

Question: Which real catalog entries lack authoritative stats, weapon ranges or legal source/race mapping?

Outside current scenarios · high confidence: The 32 prior modeled numerical reference IDs are unchanged in 70291. This is a bounded overlap result, not proof that changed dungeon rewards, Desolace equipment, Chillknife or other actual loot are integrated or globally inferior.

Current implementation / scenario: Unresolved real-item sources remain excluded from eligible selection. New dungeon/quest rarity and stat changes require catalog/acquisition reconciliation; the prior modeled numerical references are not newly optimized loot.

Impact: Manual real-gear use and a future actual-loot benchmark.

Remaining scope: The four Shaman-only Spiritcaller masks and two reviewed quest items are integrated under DATA-016. Unlisted acquisition holds, announced future level-35 equipment changes, temporary Chillknife duration/eligibility and final loot selection remain separate.

Implementation / model follow-up: Keep modeled projections distinct from obtainable items. Record source/rarity/equip/quest timing and do not promote a promised hotfix or static record to current acquisition proof.

Optional / conditional follow-up

Sources / code
Requires: Current vendor exports, acquisition records and item/effect tables for a specific excluded item.
Observe / verify: Resolve stats, weapon range, availability and unique/faction restrictions independently before making it selectable.

docs/forever_gear_data.md · assets/db_inputs/forever_gear_catalog.json · docs/modelled_gear.md · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md

Earlier references: T20 (gear/source coverage)

Related: DATA-016

Link to DATA-003

DATA-004 Ordinary equipped-set proc coverage Outside current scenarios · medium

Question: Which disabled set thresholds have enough evidence to implement?

Outside current scenarios · high confidence: Current projected items have no ordinary set membership. The unsupported ordinary thresholds are intentionally blocked, while supported passive thresholds use actual equipped piece counts. These omitted procs do not affect the forced role Tier 1 scenario, which is a separate mechanism.

Current implementation / scenario: Confirmed passive stat thresholds apply at actual piece counts; 44 unsupported thresholds remain omitted. Legacy handlers do not automatically enable catalog effects.

Impact: Manual real-item sets; synthetic ranking gear has no ordinary set bonuses.

Remaining scope: Per-threshold implementation work if ordinary real sets are enabled.

Implementation / model follow-up: Do not enable a legacy handler merely because a set name matches. Maintain threshold-level supported/unsupported records and leave the forced Tier override independent.

Optional / conditional follow-up

Sources / code
Requires: A specific ordinary set selected for support and its complete effect/trigger chain.
Observe / verify: Separate sourced passive amounts from unknown proc conditions, then check exact piece counts and duplicate suppression.

sim/core/forever_equipment_sets.go · docs/weekly_review.md · docs/forever_gear_data.md · sim/core/forever_equipment_sets_generated.go · docs/forever_tier1.md

Link to DATA-004

DATA-009 Two Arcanums of Rapidity Outside current scenarios · medium

Question: Do identical copies stack additively or multiplicatively?

Outside current scenarios · high confidence: No benchmark profile uses two Arcanums of Rapidity. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Each grants attack speed only; identical copies add. Neither is general Energy haste.

Impact: Melee/ranged swing rates with two copies.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

docs/forever_gear_data.md · sim/common/forever_enchants.go

Link to DATA-009

DATA-010 Raw haste rating conversion Outside current scenarios · medium

Question: Can future raw haste-rating items be represented without confusing ratings and flat percentages?

Outside current scenarios · high confidence: No available item has raw haste rating. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Current item/enchant representation handles sourced flat effects; the inherited raw-rating constant is not certified for new Forever rating-bearing items.

Impact: Future item ingestion and hypothetical gear sensitivity; no selected raw haste-rating item.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Separate raw-rating and flat-percent fixtures at level 60.
Observe / verify: Ten raw rating must yield 1%; an explicit 1% effect must remain 1%. Check native, web and export/import paths and their affected timers.

docs/speed_audit.md · assets/db_inputs/forever_combat_ratings.json

Link to DATA-010

DATA-011 Item-picker shared-crit EP Outside current scenarios · medium

Question: Does the gear picker value generic crit for casters the way the engine does?

Outside current scenarios · high confidence: Affects only the gear picker's displayed sort value; simulations do not use it. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Engine sharing works, but the picker can dot-product the unprojected MeleeCrit field against zero caster weight. This is a display/valuation limitation, not missing simulated crit.

Impact: Manual item sorting; native coordinate simulations are not driven by that displayed EP.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Generic-crit equipment with only a SpellCrit EP weight, plus ranged-only scope control.
Observe / verify: Compare predicted picker deltas to the projected equipment stats. Check that sharing is neither omitted nor applied twice.

docs/crit_model.md · ui/core/player.ts · sim/core/ruleset.go

Link to DATA-011

DATA-012 Explicitly scoped crit items Resolved · medium

Question: Are any remaining catalog items incorrectly treated as generic crit despite a scoped effect?

Source-backed answer · high confidence: The active scoped exception is already explicit: SAF-T enchant 8720 routes spell 1310308's bow/gun/crossbow-restricted 2% crit to ranged attacks, rather than generic equipment crit. Current projected items use intentionally generic stat fields. The earlier real-item audit is not evidence for every future proc, but no unresolved generic/scoped contradiction was found in these current flat-stat loadouts.

Current implementation / scenario: Generic equipment fields are shared; checked current flat-crit loadouts had no contradictory scoped record. That does not certify every legacy proc/enchant.

Impact: Unreviewed manual gear choices.

Remaining scope: Reopen only for a particular imported effect with a contradictory mask; legacy proc coverage stays under DATA-006.

Implementation / model follow-up: No broad crit rewrite is warranted. Fix the separate picker projection without widening ability, talent, school or ranged-only bonuses.

docs/crit_model.md · docs/forever_gear_data.md · sim/common/forever_enchants.go · tools/database/enchant_overrides.go

Related: DATA-011 · DATA-006

Link to DATA-012

DATA-013 Spell acquisition and source-scanner coverage Outside current scenarios · medium

Question: Which retained records are learnable abilities rather than NPC, rune, teaching or obsolete aliases?

Outside current scenarios · high confidence: Concerns healing spells and acquisition coverage, not the simulated DPS or tank spellbooks. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Static learned/acquisition records remain separate from runtime registration. Changed early damage ranks and all four damaging Penance ranks are registered; Lesser Heal/Heal and helpful Penance are not. Higher books/trainer/drop availability remains qualified.

Impact: Prevents missing real abilities or enabling unavailable ones.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Current SkillLineAbility/trait/teaching records plus legal runtime contexts for each supported class, tank and selectable pet.
Observe / verify: Reconcile each learned rank and alias against registration; do not treat a source-code number match as full ability coverage. Seek an actual acquisition chain before promoting held records.

docs/spell_coverage.md · artifacts/spell_coverage.json · sim/spell_sources_test.go · assets/spell_coverage_dispositions.json · artifacts/question_review/scenario_data_evidence.json · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · tools/forever_bench/ranks_october9_test.go

Earlier references: T50 in check_dispositions.md (acquisition watch) / T05 / H30

Related: HUN-007 · SCEN-008 · PRI-005

Link to DATA-013

DATA-015 Revelation explicit proc-law model Outside current scenarios · medium

Question: What is Revelation's probability law and exact eligible-event/charge behavior?

Outside current scenarios · high confidence: No benchmark profile uses the Revelation enchant. (A beta tester measured about 9.7% procs, 58 of 600, at 6% crit, against a client value of 11.) Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Sourced payload/form eligibility and an explicit profile-owned provisional proc model are integrated locally, not released. Only exact seven class-family whitelists qualify; five represented channels use a Revelation-only source-qualified child bridge. No all-spells eligibility or universal hidden chance.

Impact: Conditional enchant choice and class proc/crit value; no final selection yet.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Current beta · level 30
Requires: Obtainable Revelation enchant, recorded character crit, separated direct/periodic/noncrit/crit/miss samples, no changing other procs.
Observe / verify: Estimate eligibility and conditional chance, charge expiry/consumption and form retention; vary crit independently.
Sources / code
Requires: Final explicit model and imported/exported profile.
Observe / verify: Verify validation, probability convention, eligible family/miss boundaries and native/web roundtrip before marking integrated.

https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/common/enchant_effects.go · proto/common.proto · proto/api.proto · sim/common/forever_enchants.go · tools/forever_bench/revelation_october9_test.go

Related: DATA-006 · DRU-011

Link to DATA-015

DATA-016 70291 real-item class restrictions and catalog integration Resolved · medium

Question: Are the new actual quest items and changed Spiritcaller class restrictions represented in native/UI catalogs and imports?

Source-backed answer · high confidence: All 153 changed/21 added item leaves are adjudicated. Final compiler/DB generation and all four actual native class-restriction controls pass; ordinary UI imports are covered separately. Seven compiled price changes are noncombat. This resolves the identified catalog/class-import implementation, not live acquisition for every unlisted item.

Current implementation / scenario: The 70291 source overlay, compiled class allowlists and native/UI import restrictions are integrated. Four Spiritcaller pieces 276538–276541 are Shaman-only. Two reviewed quest items 251485/251486 are included, distinct from the unchanged synthetic budget.

Impact: Legal real-item selection/imports and conditional early quest gear, not a new synthetic budget or benchmark optimum.

Remaining scope: Acquisition holds for 38 unlisted combat changes, promised future equip-level hotfixes, temporary Chillknife source follow-up and final real-loot selection. New-item source stats are not a global-optimum claim.

Implementation / model follow-up: Preserve source masks through catalog compilation and both import paths. Keep the two reviewed real items and2799-item final real catalog separate from889 synthetic projections; final201 profiles remain modeled-v2 only.

Optional / conditional follow-up

Sources / code
Requires: Current ItemSparse masks, compiled item overlay and native/UI equipment imports.
Observe / verify: Accept each Spiritcaller piece for Shaman and reject forbidden classes; verify source-derived new-item stats and catalog count without altering synthetic budget.
Current beta · level 30
Requires: Actual obtainable quest rewards and legal class/level access, including the level-24 knife.
Observe / verify: Confirm tooltip/equip restrictions and acquisition; a modeled record is not proof of a live quest source.

https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · assets/db_inputs/forever_item_updates_70291.json · sim/core/database.go · https://wago.tools/db2/ItemSparse?build=1.60.1.70291

Related: DATA-003 · DATA-013

Link to DATA-016

DATA-005 Nature's Bounty proc rate and form restrictions Outside current scenarios · low

Question: Is the retained 2% fallback rate correct, and are triggers restricted by active form?

Outside current scenarios · high confidence: Nature's Bounty is not equipped or activated by the current synthetic loadouts. The retained fallback's 2% rate is an assumption, and checking whether a Druid has a resource bar does not establish an active-form restriction. Neither is evidence of the live set's behavior.

Current implementation / scenario: Legacy fallback has separate resource/event routes and Energy ticks over five seconds. Its 2% rate is assumed; catalog proc thresholds remain disabled.

Impact: Conditional dungeon-set resource behavior, not synthetic rankings.

Remaining scope: A dormant real-set handler; probability and form routing are unresolved if this set is brought into scope.

Implementation / model follow-up: Keep its catalog proc threshold disabled. Resource tick timing already has client support and need not be conflated with the unknown chance.

sim/common/item_sets/natures_bounty.go · docs/weekly_review.md · sim/core/forever_equipment_sets.go

Link to DATA-005

DATA-007 Item guardian stats and attacks Outside current scenarios · low

Question: What are the actual stats, attacks and resource limits of item-summoned guardians?

Outside current scenarios · high confidence: Eskhandar, the item Core Hound and Dragon's Call Whelp are not summoned by the current fixed-stat projected gear. Their approximate stats cannot affect this benchmark. This does not close the separate, active Summon Hawk talent question.

Current implementation / scenario: Eskhandar, Core Hound and Emerald Whelp retain approximate inherited stats/timing. Dragon's Call's flat-damage upstream change was not blindly imported.

Impact: Optional real-item use; absent from projected equipment.

Remaining scope: Item-specific guardian research if the corresponding real items become legal selected candidates.

Implementation / model follow-up: Do not merge item guardians with Hunter/Warlock pets or infer all guardians from one summon.

sim/common/guardians/eskhandar.go · sim/common/guardians/core_hound.go · sim/common/guardians/emerald_dragon_whelp.go · docs/mythicsim_review.md · docs/modelled_gear.md

Earlier references: M10

Related: HUN-004

Link to DATA-007

DATA-008 Unimplemented vendor item effects Outside current scenarios · low

Question: What prevents Adaptive Combat Assistant, Defender's Grip Stabilizer and Totem of the Storm from being used?

Outside current scenarios · high confidence: No benchmark profile uses these vendor items. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Absorb/explosion, location-dependent modifier and Maelstrom-related proc are excluded rather than supplied as free stats/effects.

Impact: Future real-item choices; no current modeled-item effect.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Items 272437/272440/272432, their use/equip records and spells 1291097/1291099/1291105/1291078.
Observe / verify: Implement only explicitly supported branches, preserving item cooldowns and shared use categories.
Beyond level 30 / launch
Requires: Level-60 vendor items; controlled damage into Assistant, and Maelstrom-talented Shaman with the Totem.
Observe / verify: Separate shield depletion, normal expiry and dispel; count explosion events. For Totem, compare Lightning Bolt proc rate with independently measured melee eligibility, holding talent rank fixed.

docs/forever_gear_data.md · assets/db_inputs/forever_gear_catalog.json · artifacts/question_review/scenario_data_evidence.json · https://www.wowhead.com/forever/item=272437/adaptive-combat-assistant

Related: SHA-001 · SCEN-005

Link to DATA-008

DATA-014 Talent-tree source conflicts Resolved · low

Question: Which old client/crawl disagreements still affect the legal tree rather than only historical metadata?

Source-backed answer · high confidence: Officialpost5 and70291 TraitEdge restore Impale's prerequisite. Published Warrior allocations already satisfy it. Impact3/6/9 was a Classic-prior extrapolation, not full-rank current primary data; active tree/curve correction is separate from a live tooltip hotfix claim.

Current implementation / scenario: Selectable source trees include restored full Deep Wounds→Impale, Natural Instinct naming/healing and current Impact3/7/10 curve. Class wire identities and existing legal allocations are retained.

Impact: Future tree updates and under-reviewed rank values.

Remaining scope: Current generator-input reconciliation must preserve active corrected curves; higher-rank availability and unseen hotfix overrides remain separate.

Implementation / model follow-up: Apply sourced prerequisite validation without forcing a new51-point allocation; do not regress active source metadata from retired extraction priors.

docs/talent-check-2026-09-17.md · docs/talent-values-2026-09-17.md · assets/confirmed_talents.json · ui/core/talents/trees/warrior.json · ui/core/talents/trees/shaman.json · ui/core/talents/trees/druid.json · ui/core/talents/trees/rogue.json · ui/core/talents/trees/hunter.json · ui/core/talents/trees/warlock.json · sim/core/forever_talent_tree_test.go · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-september-24/2360696 · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · tools/forever_bench/october9_patch_test.go · tools/forever_bench/mage_impact_october9_test.go

Link to DATA-014

DATA-017 Client 1.60.1.70425 item and data changes Resolved · low

Question: Which client data changes reach simulated items?

Source-backed answer · high confidence: Client spell effects and durations for 17504, 17152, 18796, 21142; 1301706.

Current implementation / scenario: Bloodrazor (five 31-damage ticks over 15 s), Destiny (12 s), Fiery War Axe (231-293 plus 20 a tick), Sulfuras immolation (7). The 22-to-674 armor aura migration keeps every value. Priest Tier 1 five-piece still removes 60 s from Devouring Plague, whose cooldown is now 24 s, so it removes all of it.

Impact: Real-item gear choices; Shadow.

Remaining scope: Whether the Tier 1 bonus will be retuned for the 24 s cooldown.

Implementation / model follow-up: sim/common/item_effects.go.

https://wago.tools/builds-diff?to=1.60.1.70425&from=1.60.1.70374

Link to DATA-017

Encounters and simulator coverage 4 active / 15 total

SCEN-001 Tank boss damage and external healing Model assumption · high

Question: What incoming damage and healing should define a representative tank comparison?

Scenario decision · high confidence: The incoming-damage and healing values are explicit controlled workloads, not claims about a known raid boss. The single workload is 300 seconds with one 3000-base/2-second attacker and 1500 external HPS; the multi workload is 90 seconds with three 1500-base/2-second attackers and 2500 HPS. There is no unique game-mechanics experiment that can prove these are the representative encounter.

Current implementation / scenario: Explicit stress scenarios: one 300-second boss or three 90-second attackers with fixed swing/AP/armor/healing settings. These are not observed raid bosses or healer behavior.

Impact: Tank DPS, rage, mana, threat and death rates; cannot be silently equated with un-attacked DPS rows.

Remaining scope: Which workloads are useful comparisons, and future measured encounter inputs.

Implementation / model follow-up: Keep exact requests and per-target threat/survival outputs. Older tables in the original tank document describe superseded models; current request artifacts are authoritative.

Answer routes

Sources / code
Requires: A chosen workload or credible raid-log encounter specification.
Observe / verify: Change damage/healing assumptions as a separately labeled scenario and compare matched inputs, rather than silently redefining the tank benchmark.

docs/tank_benchmark.md · tools/forever_tanks/main.go · artifacts/tank_support/validation.json · artifacts/protection_sunder/validation.json

Link to SCEN-001

SCEN-003 Tank target selection and threat coverage Implementation gap · high

Question: Can a tank maintain threat on the weakest-covered enemy rather than repeatedly attacking the first?

Known implementation work · high confidence: This is an APL capability gap, not unknown game behavior. The available unit references do not provide a lowest-threat target selector. Saved multi-target tank results demonstrate why total TPS cannot substitute for the weakest add's threat.

Current implementation / scenario: Existing Warrior multi-target presets remain single-target focused; summed TPS masks very low threat on an add. No lowest-threat target selector is provided.

Impact: First-class AoE tank support and meaningful optimization.

Remaining scope: Explicit target selection/cycling with target-local debuffs and threat coverage.

Implementation / model follow-up: Define a selector policy and use target-specific state; preserve self-maintained Sunder on the intended target. Do not infer raid aggro control solely from one player's total threat.

Answer routes

Sources / code
Requires: A three-target fixture with deliberately unequal per-target threat and debuff durations.
Observe / verify: Check selection, switching, maintenance and minimum target TPS; ensure current-target lookups do not accidentally read the first target's aura.

docs/tank_benchmark.md · proto/apl.proto · sim/core/apl_helpers.go · artifacts/protection_sunder/validation.json

Link to SCEN-003

CORE-022 Benchmark raid buffs after the crit-aura split and raid-wide totems Model assumption · medium

Question: Which crit auras and air totems should the reference raid supply?

Credible working answer · moderate confidence: Before 1.60.1.70425 one druid aura supplied 3% crit to everyone and Grace of Air was exclusive with Windfury. Both constraints are gone and the buffs reach the whole raid, so a raid with one Feral and one Balance Druid and two air Shamans supplies all of them.

Current implementation / scenario: Both Leader of the Pack and Moonkin Aura for every profile; melee groups get Windfury and Grace of Air, Hunters Grace of Air, casters neither. Tank support is unchanged.

Impact: All melee DPS rows (+3% crit returns from Leader of the Pack; Grace of Air adds Agility).

Remaining scope: Whether raids will actually field both Druid specs and two air Shamans.

Implementation / model follow-up: ForeverRaidBuffs carries Leader of the Pack; tools/forever_bench/main.go keeps Grace of Air for melee and Hunters.

https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · sim/core/test_utils.go · tools/forever_bench/main.go

Link to CORE-022

SCEN-005 General absorb-pool and shielding threat Needs evidence · medium

Question: Do generic shields consume finite incoming damage and generate correct threat?

Partly answered · moderate confidence: Core partial/full/replacement/expiry callback controls pass. Dry probes do not consume shields. Incoming Rage restores absorbed post-outcome damage; callback wiring does not prove shield-threat or replacement scripts.

Current implementation / scenario: Finite after-mitigation absorb delivery and metrics remain; actual incoming context now supports complete-depletion callbacks. SoF uses one explicitly provisional replacement pool/lifetime, not accumulated rank-specific shields.

Impact: Priest/healing shields and future Seal of Fury integration; not a reason to revert correct Bulwark behavior.

Remaining scope: General shielding threat and pool overlap priority; application-order convention, SoF script lifetime/replacement and broader school-filtered/heal-engine coverage.

Implementation / model follow-up: Keep generated/absorbed/unused separate, with declared provisional pool lifetime0/5/30/60 comparisons recorded. Do not refund on expiry/replacement or count unused capacity as healing.

Answer routes

Sources / code
Requires: A shield with a sourced amount/duration and sequential controlled incoming hits.
Observe / verify: Verify partial and full consumption, expiry, overwrites and ordering without converting unconsumed capacity into effective healing.
Current beta · level 30
Requires: Priest with Power Word: Shield, a cooperating player and a controlled hostile target with threat observation.
Observe / verify: Separate threat on shield application, actual absorption and unused expiry; do not attribute incidental healing or damage threat to the shield.

sim/core/shield.go · sim/paladin/templars_bulwark.go · docs/weekly_review.md · sim/core/spell_result.go · proto/api.proto · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/core/shield_depletion_test.go · sim/core/forever_incoming_rage.go

Earlier references: H20 / M02 (remaining generic-shield portion)

Related: PAL-006 · WAR-003

Link to SCEN-005

SCEN-016 Overlapping seeds in historical validation Resolved · high

Question: Which earlier pooled comparisons counted overlapping iteration seeds as independent replications?

Source-backed answer · moderate confidence: The overlap concerned historical validation runs. Current releases draw every matrix, gear search and tank check from separate, well-spaced seed ranges, so the published means are not affected.

Current implementation / scenario: New validation guards compare complete iteration-seed intervals, not only distinct starting seeds. Historical overlapping cohorts retain raw means and requests but must not be called independent; conservative interval qualification is implemented locally.

Impact: Legacy pooled sample-size and uncertainty claims need reassessment; this does not by itself invalidate saved DPS means or change combat behavior.

Remaining scope: None.

Implementation / model follow-up: Keep separate seed ranges for each release step.

Optional / conditional follow-up

Sources / code
Requires: Saved requests/iteration counts and the explicit seed-range inventory.
Observe / verify: Reject overlap within claimed independent replications. Recalculate uncertainty from unique iterations where available, or repeat only material decisions with disjoint ranges; do not rewrite archived raw evidence.

sim/core/sim.go · docs/protection_sunder.md · artifacts/tank_support/validation.json · artifacts/protection_queue/validation.json · tools/forever_bench/validate_gear.py · artifacts/question_review/seed_ranges.json · artifacts/protection_sunder/validation.json · tools/forever_bench/seed_intervals.py · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md

Related: SCEN-004

Link to SCEN-016

SCEN-002 Non-tank external healing abstraction Outside current scenarios · medium

Question: How should health costs constrain Life Tap, Death and self-damaging consumables in DPS scenarios?

Outside current scenarios · high confidence: Healing of non-tanks does not change DPS results. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Self-damage is excluded from enemy DPS. Non-tanking resource use retains its external-healing abstraction rather than a finite healer budget; tank requests pay real health costs.

Impact: Resource sustainability and practical healer burden.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: An explicit incoming-damage/healer-resource model and failure condition for non-tanks.
Observe / verify: Compare paid health, effective healing, overheal and deaths separately from outgoing enemy damage.

sim/warlock/lifetap.go · docs/upstream_elliot_review.md · docs/forever-70009.md · sim/priest/shadow_word_death.go

Link to SCEN-002

SCEN-006 Blood Pact talent bonus and pet-provider lifecycle Resolved · medium

Question: Does the current Improved Imp talent still increase Blood Pact, and do owned pet transitions update party buffs?

Source-backed answer · high confidence: Forever Blood Pact uses sourced base Stamina without the inherited talent bonus. Owned Imp support follows pet enable/disable, sacrifice and reset; external support remains independent, party-scoped and non-duplicating. Native lifecycle and initial-health checks pass.

Current implementation / scenario: Forever Blood Pact uses sourced base Stamina without the inherited talent bonus. Owned Imp support follows pet enable/disable, sacrifice and reset; external support remains independent, party-scoped and non-duplicating. Native lifecycle and initial-health checks pass.

Impact: Tank health and party simulations involving an Imp.

Remaining scope: None for the resolved question. Current201-profile results/native-web defaults are refreshed; actual publication is separately evidenced by deploy.yml and live resource checks.

Implementation / model follow-up: Corrected locally with focused native controls; affected exact profiles and published results require refresh before these changes are released.

docs/weekly_review.md · sim/warlock/warlock.go · sim/core/buffs.go · tools/forever_bench/blood_pact_test.go · ui/core/components/inputs/buffs_debuffs.ts · https://www.wowhead.com/forever/spell=18694/improved-imp · artifacts/question_review/class_evidence.json · sim/warlock/imp.go

Related: WL-006 · WL-008

Link to SCEN-006

SCEN-008 In-range and reactive damage omissions Outside current scenarios · medium

Question: Which supported damage actions remain missing outside the stationary reference rotation?

Outside current scenarios · high confidence: These actions matter only for AoE, solo or alternate targets, not the single-target benchmark. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Hunter Concussive Shot, Disengage and Frost Trap initial-control paths plus Mage Impact are now registered. Hellfire/Chastise and other encounter-dependent actions remain absent; Freezing Trap actual freeze/break payload and geometry are not fully implemented.

Impact: AoE, solo and alternate target types, not proof of a current five-minute gain.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

Optional / conditional follow-up

Sources / code
Requires: Current Hellfire/Chastise effect chains and explicit in-range or Humanoid targets.
Observe / verify: Verify enemy versus self-damage, rank availability, target restrictions and range; the present Dragonkin/20-yard controls must not gain free damage.

docs/spell_coverage.md · artifacts/spell_coverage.json · docs/upstream_elliot_review.md · assets/spell_coverage_dispositions.json · sim/mage/frost_area.go · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/hunter/utility.go · sim/core/crowd_control.go · tools/forever_bench/hunter_utility_october9_test.go

Related: DATA-013 · CORE-017 · MAG-008

Link to SCEN-008

SCEN-010 Forced Tier 1 and synthetic stat scaling Outside current scenarios · medium

Question: What do the Tier-off and +10%/+50% scenarios establish?

Outside current scenarios · high confidence: Defines the sensitivity columns; it does not change the simulated results themselves. Moved outside current scenarios; it can be reopened if a scenario starts to depend on it.

Current implementation / scenario: Full role Tier 1 is forced independently of equipment. Stat/weapon scaling holds enchants/procs/talents/APLs fixed; amplification is finite-range curvature, not proof of exponential growth or future loot.

Impact: Interpretation of all matrix gain columns.

Remaining scope: None for current scenarios.

Implementation / model follow-up: No change.

docs/forever_tier1.md · docs/modelled_gear.md · tools/forever_bench/sensitivity.py · sim/core/equipment_scaling_test.go

Link to SCEN-010

SCEN-011 Windfury Totem in animal forms Resolved · medium

Question: Does the redesigned party buff proc for Cat and Bear?

Source-backed answer · high confidence: Blizzard explicitly supports Windfury Totem benefits in Cat/Bear forms. Form eligibility is implemented; this does not establish every proc amount/charge rule, which remains under PAL-003.

Current implementation / scenario: Blizzard explicitly supports Windfury Totem benefits in Cat/Bear forms. Form eligibility is implemented; this does not establish every proc amount/charge rule, which remains under PAL-003.

Impact: Feral and Bear damage, rage and proc interactions.

Remaining scope: None for the resolved question. Current201-profile results/native-web defaults are refreshed; actual publication is separately evidenced by deploy.yml and live resource checks.

Implementation / model follow-up: Corrected locally with focused native controls; affected exact profiles and published results require refresh before these changes are released.

docs/forever-70009.md · sim/core/buffs.go · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-september-24/2360696 · https://news.blizzard.com/en-us/article/24301515/world-of-warcraft-forever-class-deep-dives-hunter-and-druid

Earlier references: T52 in in_game_checks.md (Totem/form eligibility)

Related: PAL-003 · DRU-011

Link to SCEN-011

SCEN-014 Legacy regression-suite alignment Resolved · medium

Question: Can the full inherited test suite run against the current Forever data and presets, beyond the maintained release checks?

Source-backed answer · high confidence: Five reproduced legacy metadata checks now pass: current sourced rank fixtures, declaration-order talent mapping, legal selectable presets, the review-portal landing redirect and current fallback-menu names. The same check uncovered and corrected the real Hunter Careful Aim/Lethal Attacks decode swap. This is not a claim that every inherited suite or unsupported healer scenario is complete.

Current implementation / scenario: Five reproduced legacy metadata checks now pass: current sourced rank fixtures, declaration-order talent mapping, legal selectable presets, the review-portal landing redirect and current fallback-menu names. The same check uncovered and corrected the real Hunter Careful Aim/Lethal Attacks decode swap. This is not a claim that every inherited suite or unsupported healer scenario is complete.

Impact: Automated regression coverage; a failing legacy fixture is not itself evidence that current game calculations are wrong.

Remaining scope: None for the resolved question. Current201-profile results/native-web defaults are refreshed; actual publication is separately evidenced by deploy.yml and live resource checks.

Implementation / model follow-up: Corrected locally with focused native controls; affected exact profiles and published results require refresh before these changes are released.

.github/workflows/deploy.yml · sim/confirmed_talents_test.go · sim/talents_test.go · sim/core/simsignals/api_test.go · sim/paladin/paladin.go · ui/elemental_shaman/presets.ts · artifacts/question_review/scenario_data_evidence.json · proto/hunter.proto · ui/core/community_builds.ts · ui/holy_paladin/presets.ts

Related: DATA-014

Link to SCEN-014

SCEN-007 Healing specializations and utility spellbooks Outside current scenarios · low

Question: Which healing/control-only talents are outside current DPS and tank coverage?

Outside current scenarios · high confidence: The overnight source review records a concrete healing-family feature gap under PRI-005. This is implementation coverage, not a request to infer an unpublished DPS multiplier.

Current implementation / scenario: Restoration agents and much of Priest/Paladin healing remain incomplete. Captured70291 Lesser Heal/Heal and Penance healing rows are not registered spells; new Intellect healing does not invent a healer rotation.

Impact: Healer simulation and encounters with CC/dispels, not current stationary DPS.

Remaining scope: Future supported healer agents/spellbooks and the identified registrations; current PvE DPS/tank results do not constitute healer validation.

Implementation / model follow-up: Keep the capabilities boundary and concrete PRI-005 backlog visible; avoid an unrequested broad healing engine.

docs/beta-pass/druid.md · docs/beta-pass/paladin.md · docs/beta-pass/priest.md · docs/beta-pass/shaman.md · docs/forever_tier1.md · docs/tank_benchmark.md · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md

Related: PRI-005

Link to SCEN-007

SCEN-009 Geometry, movement, kills and latency Outside current scenarios · low

Question: Which real fight events are absent from the stationary benchmark?

Outside current scenarios · high confidence: Terrain, movement paths, add deaths and human latency are deliberately absent from the stationary reference workloads. Their exact real-world behavior cannot alter a run containing none of those events. This does not make Consecration damage or current three-target threat irrelevant.

Current implementation / scenario: No terrain/vertical targeting, realistic movement path, encounter add deaths or general player latency. AoE uses target/range approximations; remainingTime is encounter time, not per-target TTK.

Impact: Consecration placement, cone targeting, grenades, kill procs, stealth/control and short-lived adds.

Remaining scope: These become requirements for a moving/add-lifetime encounter, not current coefficient tests.

Implementation / model follow-up: Keep the encounter scope visible and do not interpret encounter remainingTime as measured target TTK.

docs/forever-70009.md · sim/mage/frost_area.go · sim/core/apl_values_encounter.go · docs/beta-pass/hunter.md · docs/tank_benchmark.md

Link to SCEN-009

SCEN-012 Other omitted mitigation and threat effects Outside current scenarios · low

Question: Which acknowledged non-DPS effects become relevant with incoming attacks?

Outside current scenarios · high confidence: A taunt debuff is not highest-threat equalization or forced AI targeting. Specific Hunter/Mage CC additions do not supply a complete PvP/raid encounter engine. Screech AP reduction is already implemented under HUN-009, not a remaining omission.

Current implementation / scenario: No complete taunt/aggro target-selection engine. SoF judgement carries a sourced four-second taunt aura only; explicitly eligible CC/DR now exists but default scenarios remain immune/off. Other omitted utility remains conditional.

Impact: Conditional tank/raid/solo scenarios; not universal current benchmark damage errors.

Remaining scope: Threat-driven target swaps, general interrupt/fear/sleep/charm/knockback and applicable encounter utility.

Implementation / model follow-up: Retain explicit per-effect capability boundaries instead of claiming all utility or all taunts are simulated.

docs/beta-pass/druid.md · sim/mage/counterspell.go · sim/core/target.go · docs/tank_benchmark.md · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · sim/paladin/sof.go · sim/core/crowd_control.go

Related: HUN-009 · SCEN-003 · ROG-006 · CORE-017

Link to SCEN-012

SCEN-013 Power Infusion on another raid member Resolved · low

Question: Can an owned Priest cast Power Infusion on a selected other player?

Source-backed answer · high confidence: Owned Power Infusion can target a selected friendly unit. Omitted target defaults to self; explicitly unassigned disables it. The magic-damage/healing bonus and mana reduction do not stack with another PI provider. Native target and overlap checks pass.

Current implementation / scenario: Owned Power Infusion can target a selected friendly unit. Omitted target defaults to self; explicitly unassigned disables it. The magic-damage/healing bonus and mana reduction do not stack with another PI provider. Native target and overlap checks pass.

Impact: Multi-player raid support and support-DPS comparisons.

Remaining scope: None for the resolved question. Current201-profile results/native-web defaults are refreshed; actual publication is separately evidenced by deploy.yml and live resource checks.

Implementation / model follow-up: Corrected locally with focused native controls; affected exact profiles and published results require refresh before these changes are released.

sim/priest/power_infusion.go · docs/beta-pass/priest.md · sim/core/buffs.go

Link to SCEN-013

Resolved questions and superseded claims 0 active / 20 total

DONE-001 Healing is not inferred spell damage Resolved · high

Question: Are pure healing effects accidentally converted to spell damage?

Previously resolved · high confidence: The generic healing-to-damage fallback is absent from current equipment compilation. Hybrid healing/damage effects retain separate sourced values, including enchants and supported set effects.

Current implementation / scenario: The generic one-third fallback is removed. Sourced hybrid healing/damage effects are imported separately, including enchants.

Impact: Historical gear/enchant misvaluation; current projected catalogs and real-effect imports retain explicit values.

Remaining scope: Only item-specific contradictory evidence would reopen this question.

Implementation / model follow-up: No blanket conversion or stripping of legitimate hybrid bonuses is warranted.

docs/forever_gear_data.md · docs/mechanics_review.md · tools/database/compile_forever_equipment.py · tools/database/import_forever_effect_audit.py · tools/forever_bench/enchants_test.go

Earlier references: T22 / H02

Link to DONE-001

DONE-002 Hot Streak and Clearcasting consumption Resolved · high

Question: Can the old persistent proc buffs still provide repeated accelerated/free casts?

Previously resolved · high confidence: Hot Streak is consumed by completed Pyroblast and Clearcasting consumption checks underlying paid cost. The old permanent-proc defects are not present in current code.

Current implementation / scenario: Completed Pyroblast consumes Hot Streak; eligible paid damage casts consume Clearcasting using underlying cost, not the discounted zero.

Impact: Historical Mage resource/damage error fixed; edge timing has separate active entries.

Remaining scope: Proc ordering and in-flight edge cases are narrower MAG-004/MAG-005 questions, not the same fixed bug.

Implementation / model follow-up: Keep the existing consumption regressions; do not preserve Hot Streak indefinitely to increase Fire damage.

docs/history_review.md · docs/mechanics_review.md · sim/mage/talents.go · tools/forever_bench/mage_test.go

Earlier references: H01 / H03 / T23 / T34

Related: MAG-004 · MAG-005

Link to DONE-002

DONE-003 Death, Early Demise and Penance timing Resolved · high

Question: Are these Priest abilities/talents still missing or using evenly spaced Penance bolts?

Previously resolved · high confidence: The existing two-second timing was already correct.70291 changes numbers and rank coverage, not an extra channel-speed buff; maximum92+.19SP/bolt and385Mana are tested.

Current implementation / scenario: Four learned Death ranks/Early Demise remain. Four damaging Penance ranks now use150/220/270/385Mana and0.19SP per bolt; immediate0/1/2-second channel timing remains unchanged. Helpful Penance is not registered.

Impact: Old spellbook/mana/execute omissions corrected; Death script remains separate.

Remaining scope: Healing feature gap under PRI-005 and higher-rank acquisition are not covered by timing validation.

Implementation / model follow-up: Retain resolved timing and source numbers; current201-profile benchmark/native-web exports are refreshed. Publication receipt is separately linked to deploy.yml.

docs/spell_coverage.md · docs/build_updates.md · sim/priest/shadow_word_death.go · sim/priest/talents.go · sim/priest/penance.go · tools/forever_bench/priest_death_test.go · tools/forever_bench/channel_haste_test.go · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · tools/forever_bench/ranks_october9_test.go

Earlier references: T19 / H10 / H11

Related: PRI-001 · PRI-005

Link to DONE-003

DONE-004 Mage Armor, Innervate accounting and timing Resolved · high

Question: Are Mage Armor and apparent negative Innervate returns still broken?

Previously resolved · high confidence: Current Mage Armor uses its supported casting-regeneration effect. Innervate's extra regeneration is separately accounted, and its external activation threshold considers expected regeneration/missing mana rather than a fixed 85% gate.

Current implementation / scenario: Current Mage profiles use available Mage Armor. Innervate bonus regeneration has its own accounting; optional external Innervate waits for enough missing mana rather than firing at a fixed 85%.

Impact: Historical mana/reporting issues; the ranked Mages have no external Innervate.

Remaining scope: The optional MP5 interpretation and any future source-specific regeneration correction are separate questions.

Implementation / model follow-up: Preserve the repaired default mode; an assumed fivefold MP5 mode is not required to fix Innervate.

docs/mechanics_review.md · docs/build_updates.md · docs/mana_regeneration.md · sim/mage/armors.go · sim/core/buffs.go · sim/core/mana_regen_test.go · tools/forever_bench/mage_armor_test.go

Link to DONE-004

DONE-005 Shared crit aura, legal imbues and air buffs Resolved · high

Question: Are Moonkin/Leader and Windfury/Grace still incorrectly stacked?

Previously resolved · high confidence: The raid crit auras share one global benefit; incompatible air buffs are exclusive. Windfury Weapon is Shaman-only, and nonweapon/empty slots do not receive weapon imbues. The current checks cover these configuration rules. October 15 update: Client 1.60.1.70425 splits the auras again (24907 aura 57 spell crit, 24932 aura 52 weapon crit, both effect 65 raid area) and the October 15 notes make Grace of Air non-exclusive with Windfury and Flametongue Totem and all four totems raid-wide. The shared-aura and Grace/Windfury rules above were replaced accordingly.

Current implementation / scenario: Leader of the Pack is 3% melee and ranged crit and Moonkin Aura 3% spell crit; both are raid-wide and the benchmark raid carries both. Grace of Air no longer excludes Windfury or Flametongue Totem; Windfury and Flametongue still exclude each other. Windfury Weapon is Shaman-only and empty or non-weapon slots receive no imbue.

Impact: Confirmed configuration errors corrected; generic crit remains distinct from scoped crit effects.

Remaining scope: Windfury's detailed attack resolution and animal-form eligibility remain separate evidence questions.

Implementation / model follow-up: sim/core/buffs.go, sim/core/raid.go (raid-wide sharing), the web buff controls and the benchmark raid (both crit auras; melee groups take Grace of Air and Windfury).

docs/crit_model.md · docs/forever-70009.md · sim/core/equipment_validation.go · sim/core/buffs.go · sim/core/forever_consumes_test.go · tools/forever_bench/crit_aura_test.go · https://us.forums.blizzard.com/en/wow/t/wow-forever-beta-development-notes-%E2%80%93-updated-october-15/2360696/6 · https://wago.tools/builds-diff?to=1.60.1.70425&from=1.60.1.70374

Link to DONE-005

DONE-006 Legal armor, shield Block and Bulwark Resolved · high

Question: Can Feral import plate, do shields lack base Block, or is Bulwark unlimited reduction?

Previously resolved · high confidence: Druid armor ceilings are enforced natively, shield base Block is imported, active Shield Block recasts replenish charges, and Bulwark consumes a finite eight-second pool instead of granting unlimited reduction.

Current implementation / scenario: Native armor ceilings reject illegal Druid armor; intrinsic shield Block is imported; Shield Block recasts restore charges; Bulwark is a finite eight-second pool and has no GCD.

Impact: Reproduced tank/import defects are fixed, not remaining gameplay tests.

Remaining scope: The generic Shield helper and other unimplemented tank effects are separate from these corrected paths.

Implementation / model follow-up: Preserve armor legality and finite-pool regressions; do not equate the remaining generic-shield gap with a broken Bulwark.

docs/weekly_review.md · sim/paladin/templars_bulwark.go · sim/core/equipment_validation.go · sim/warrior/shield_block.go · tools/forever_bench/warrior_shields_test.go

Earlier references: W1 / W2 / W4 / W8 / H20 (Bulwark portion)

Related: SCEN-005

Link to DONE-006

DONE-007 Life Tap, Moonfury and explicit pet bonuses Resolved · high

Question: Are the latest source-backed class-effects fixes actually applied?

Previously resolved · high confidence: Life Tap uses the sourced Spirit conversion with mana-only Tier multiplication; Moonfury affects full Arcane/Nature damage; the explicit Demonic Knowledge and Focused Fire pet effects are present. These are distinct from unresolved general pet-inheritance coefficients.

Current implementation / scenario: Life Tap uses Spirit with separate health/mana accounting; Moonfury multiplies full Arcane/Nature damage; Demonic Knowledge and Focused Fire reach their explicit pet targets. General inheritance already existed.

Impact: Published fixed-profile correctness release; not pending code changes.

Remaining scope: No reversal of these supported corrections; Imp-specific new findings are tracked separately.

Implementation / model follow-up: Keep the existing class-effect regressions and provider lifecycle checks.

docs/upstream_elliot_review.md · sim/warlock/lifetap.go · sim/druid/talents.go · tools/forever_bench/upstream_followup_test.go

Earlier references: T26 (Moonfury portion) / latest upstream source-supported gaps

Link to DONE-007

PAL-014 Protection seal refresh waste Resolved · high

Question: Does Protection repeatedly pay for an already active Seal of Righteousness?

Previously resolved · high confidence: The Protection APL no longer contains the unconditional active-Seal filler. The regression and archived comparisons address a real repeated-cost defect, not an assumption that removing all seal damage is optimal.

Current implementation / scenario: The unconditional seal filler is removed. Seal maintenance, paid pre-pull buffs and mana-reserved Consecration replace the wasteful rotation in all three ranked Protection profiles.

Impact: The old five-minute rotation spent about 16,100 unnecessary mana on seal refreshes. The corrected same-gear/talent rotation gains roughly 10–11% single-target DPS.

Remaining scope: Resource-threat accounting remains CORE-014 and older comparison-seed qualification remains SCEN-016.

Implementation / model follow-up: Keep the maintenance fix; do not restore mana waste merely to reproduce old threat numbers.

docs/paladin_mana.md · ui/protection_paladin/apls/forever_protection.apl.json · tools/forever_bench/paladin_apl_test.go · sim/paladin/judgement.go

Related: CORE-014 · PAL-001

Link to PAL-014

SCEN-004 Tank selection after correctness and support corrections Resolved · high

Question: Has tank gear/talent selection been repeated under the corrected queue model and common support?

Previously resolved · high confidence: The current frozen-model bounded selection, exact source/native/web bindings and fresh independent default checks are complete. Central accepted changes are conditional on declared incoming Rage/support/SoF conventions, not empirical server verification.

Current implementation / scenario: 70291 bounded selection is complete: 10 Warrior and 3 Paladin recipes selected, all 4 Bear players retained with corrected owned Faerie Fire.152 selection workload results at two disjoint5000 windows, plus136 fresh post-source-sync default/control workload rows, pass central five point-estimate guards. Six alternate-model sensitivity cases fail; no global optimum or no-survival-loss confidence claim.

Impact: The prior search remains superseded. New conditional bounded selections and independently seeded guards are recorded; alternate-model failures and small tradeoffs prevent a global-optimum or no-survival-loss claim.

Remaining scope: Unknown server models, alternate-model guard failures and bounded local-coordinate selection limits remain explicit; workflow completion does not resolve them.

Implementation / model follow-up: Preserve complete native commands/requests/results/phase statistics and quarantined Bear phases. Do not optimize unknown script parameters, reuse historical means as current controls or call point guards confidence-proof of no survival loss.

Optional / conditional follow-up

Sources / code
Requires: An authorized selection pass with current frozen controls and survival/threat constraints.
Observe / verify: Retain only validated gains under corrected mechanics; do not compare against historical invalid controls.

docs/protection_queue.md · docs/tank_selection.md · artifacts/tanks/current/validation.json · tools/forever_bench/tank_search.go · docs/tank_support.md · artifacts/tanks/shared_support_controls.json · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md

Related: WAR-013 · PAL-005 · DRU-011 · DATA-002 · SCEN-001 · SCEN-003 · SCEN-016 · WAR-003 · PAL-006 · DATA-015

Link to SCEN-004

SCEN-015 Unequal tank raid support Resolved · high

Question: Do tank defaults provide consistent raid support and supported weapon consumables?

Previously resolved · high confidence: The inherited Bear-duty discrepancy was reproduced and fixed in the Druid native/TS support branches. All four Bear races cast 9907 in both workloads without external Faerie Fire. Warrior/Paladin support is unchanged; exact player/loadout and explicit request-support preservation are audited separately.

Current implementation / scenario: Shared native/web support covers all 17 tanks. The confirmed 70291 integration bug is corrected: all 4 Bear requests remove external Faerie Fire so their APL casts 9907; their complete players remain exact originals. All 13 Warrior/Paladin external Faerie Fire settings remain unchanged. Paladin retains Brilliant Wizard Oil, Curse of Elements and Titans.

Impact: Earlier Bear external-Faerie-Fire requests masked the owned duty. Release1 Bear study phases are quarantined; corrected original LOADOUTS preserve exact players but explicitly migrate support, not falsely identical historical REQUESTS. This plumbing fix does not verify hidden class mechanics.

Remaining scope: Scenario plumbing is resolved. Incoming-Rage normalization and other server-mechanics questions retain their separate provisional status.

Implementation / model follow-up: Retain TestOctober9BearOwnsFaerieFire and exact native/web race-profile bindings. Never retrofit external Faerie Fire into current Bear controls or call superseded Bear phases canonical.

Optional / conditional follow-up

Sources / code
Requires: All four Bear races, single and three-attacker workloads, native/web shared support helpers.
Observe / verify: Assert no external Faerie Fire and actual owned 9907 casts; preserve every player field and all other support fields.

sim/core/forever_tank_support.json · sim/core/forever_tank_support.go · ui/core/forever_tank_support.ts · tools/forever_bench/october9_patch_test.go · docs/october-8-overnight-update.md · artifacts/correctness/october_9/selection-evidence.json · docs/tank_support.md · artifacts/tank_support/validation.json

Related: SCEN-001 · SCEN-004 · SCEN-016 · SCEN-006

Link to SCEN-015

WAR-013 Protection next-swing queue bypass Resolved · high

Question: Did the Protection APL add Heroic Strike/Cleave damage without replacing a white swing?

Previously resolved · high confidence: The selected Protection actions use the queue entries, and direct APL invocation of the damage components is rejected. This fixes the demonstrated extra-damage/rage bypass without asserting that every rage formula is correct.

Current implementation / scenario: Both actions use their queue entries in all ten Protection profiles. An on-next-swing flag rejects direct APL casts of Heroic Strike, Cleave and Maul damage components. Fury and Bear already used the correct queue entries.

Impact: The old Protection DPS, rage availability and threat were inflated. Earlier selection-gain claims are invalid; current results use corrected inputs.

Remaining scope: Historical validation uncertainty is covered by SCEN-016; it does not undo the reproduced queue bug or its structural fix.

Implementation / model follow-up: Retain the corrected tags and queue guard. Do not reclassify normal queued attacks as a new rage exploit.

docs/protection_queue.md · sim/core/apl_actions_casting.go · sim/warrior/heroic_strike_cleave.go · tools/forever_bench/queued_attacks_test.go

Related: SCEN-004 · WAR-001 · WAR-003

Link to WAR-013

WAR-014 Protection Sunder maintenance starvation Resolved · high

Question: Does the Protection rotation preserve five Sunder stacks under competing rage demands?

Previously resolved · high confidence: The current APL reserves rage and gives Sunder an eight-second refresh window. The archived observer and reproduced-lapse regression support a substantial reduction in starvation, not an absolute guarantee against every avoidance/resource sequence.

Current implementation / scenario: An eight-second refresh window and dynamic-cost rage reserves substantially reduce lapses, with no free external armor debuff. Essential defenses retain priority.

Impact: In matched coverage samples, five-stack losses fall from 42.22% to 0.57% of single-attacker fights. Mean post-ramp time below five falls from 4.86 to 0.053 seconds.

Remaining scope: None for the reproduced priority defect. The residual rare lapse is disclosed rather than hidden with an external armor debuff.

Implementation / model follow-up: Keep the resolved status and the measured, nonzero residual-lapse qualification. This review's resource-threat leads do not justify silently rerunning or relabeling its archived numbers.

docs/protection_sunder.md · tools/forever_bench/sunder_maintenance_test.go · tools/sunder_audit/main.go

Related: WAR-013 · SCEN-004

Link to WAR-014

DONE-008 Demonic Sacrifice coexistence false lead Resolved · medium

Question: Can a no-Pact Warlock keep sacrifice benefits with an active pet?

Previously resolved · high confidence: The no-Pact sacrifice coexistence lead was disproved by the full reset/enable sequence and zero combat benefit. Reading only owner Reset was incomplete.

Current implementation / scenario: Pet reset/enable ordering cancels the sacrifice. Full-run checks recorded zero combat aura/mana benefit without Pact.

Impact: The earlier code-read suspicion was disproved; not an exploit to optimize.

Remaining scope: None for that alleged defect.

Implementation / model follow-up: Keep the retraction and do not add a redundant behavior-changing guard based on the old partial trace.

docs/mythicsim_review.md · sim/warlock/warlock.go · sim/warlock/pet.go

Earlier references: T09 / M13

Link to DONE-008

DONE-009 Lightning Shield's unsupported Earth Shock reset Resolved · medium

Question: Does recasting Lightning Shield reset Earth Shock?

Previously resolved · high confidence: The unsupported Lightning Shield personal-timer reset was removed. The shared shock timer had already prevented a demonstrated extra Earth Shock, so the historical finding is not evidence of a former large DPS exploit.

Current implementation / scenario: The inherited personal-timer reset is removed. The shared shock timer had already blocked demonstrated extra casts, so it was not a proven DPS exploit.

Impact: Correctness cleanup, not an unexplained damage gain.

Remaining scope: Normal Maelstrom and shield behavior have separate questions.

Implementation / model follow-up: Retain the real-cast/shared-cooldown regression and the qualified historical account.

docs/mythicsim_review.md · sim/shaman/lightning_shield.go

Earlier references: T47 / M09

Link to DONE-009

DONE-010 Seal of Command SP placement comparison Resolved · medium

Question: Does this engine omit the 70% multiplier on personal SP that upstream added?

Previously resolved · high confidence: Seal of Command's personal SP contribution already passes through its 0.7 damage multiplier. The upstream placement change therefore does not establish an omitted multiplier here. Target Holy bonus is a separate later calculation.

Current implementation / scenario: No: proc DamageMultiplier is 0.7 × Improved Seals × weapon specialization, with BonusCoefficient 0.29. Personal SP passes through that multiplier; target Holy bonus is applied separately afterward.

Impact: A recent review overstated the difference by comparing field placement rather than final math. Righteousness remains genuinely unresolved.

Remaining scope: Independent seal trigger/scaling questions do not justify multiplying personal SP by 0.7 twice.

Implementation / model follow-up: No correction to this traced multiplier-placement claim.

sim/paladin/soc.go · sim/core/spell_result.go · docs/upstream-forever-followup-2026-09-23.md · docs/upstream_elliot_review.md

Earlier references: T57 (Command placement)

Related: PAL-001 · PAL-003

Link to DONE-010

DONE-011 Cone of Cold, Frost Nova and owned Mana Tide Resolved · medium

Question: Are the later sourced spell additions still absent?

Previously resolved · high confidence: The ranked Mage area spells and owned Mana Tide cast/pulses are implemented. Mage damage checks use ten-yard proximity and do not grant boss roots/Shatter; Mana Tide replaces the relevant water effect.

Current implementation / scenario: Mage area spells have ranked damage, cooldowns and 10-yard checks. Owned Mana Tide has its real cast, costs, four pulses and water replacement. Control geometry and owned Mana Spring remain separate gaps.

Impact: Old spell-coverage pages are historical, not evidence these registrations are missing.

Remaining scope: Owned Mana Spring and movement/control geometry are distinct gaps.

Implementation / model follow-up: Do not retain these completed spell additions in the unresolved queue.

docs/upstream_elliot_review.md · sim/mage/frost_area.go · sim/shaman/water_totems.go

Related: SHA-004 · SCEN-009

Link to DONE-011

DONE-013 Self-damage is not encounter DPS Resolved · medium

Question: Do Demonic Rune and Sapper self-hits inflate outgoing DPS?

Previously resolved · high confidence: Self-target damage does not contribute to enemy DPS/threat. Demonic Rune and Sapper retain their self-cost/resource behavior under the configured health model.

Current implementation / scenario: Self-hits are excluded from outgoing encounter damage/threat while preserving damage-taken/resource behavior.

Impact: Historical accounting inflation fixed; health sustainability is a separate scenario assumption.

Remaining scope: Finite non-tank healing support is a scenario question, not an unresolved DPS aggregation bug.

Implementation / model follow-up: Preserve the enemy/self-target accounting regression and the health-support qualification.

docs/forever-70009.md · docs/weekly_review.md · sim/core/spell_result.go

Related: SCEN-002

Link to DONE-013

DONE-014 Hunter trigger scopes and shared rank cooldowns Resolved · medium

Question: Are Expose Prey and the audited per-rank cooldown bypasses still incorrect?

Previously resolved · high confidence: October 8 lengthens Expose Prey only. Separate window tests pass and generic trap spell hits still do not become eligible attacks.

Current implementation / scenario: Existing Hunter proc-mask/Mark ownership and shared rank timers remain. Expose Prey now has its own10-second opportunity; ordinary dodge-triggered Mongoose remains5 seconds.

Impact: Earlier trigger, free-cost and rank-bypass findings are implemented corrections, not open proc-rate tests.

Remaining scope: Other Hunter control/downrank questions remain under DATA-001/CORE-017.

Implementation / model follow-up: Keep the source-backed trigger boundaries and avoid extending the ordinary dodge window.

docs/mechanics_review.md · docs/history_review.md · docs/upstream_elliot_review.md · tools/forever_bench/hunter_expose_prey_test.go · https://us.forums.blizzard.com/en/wow/t/2360696/5 · docs/october-8-overnight-update.md · tools/forever_bench/druid_hunter_shaman_october9_test.go

Earlier references: T31 (Expose Prey) / H17 / H18 / H21

Related: HUN-008 · WAR-011 · DATA-001 · CORE-017

Link to DONE-014

PAL-015 Swift Judgement cooldown timing Resolved · medium

Question: Does Protection actually exploit Swift Judgement’s reset and free cast?

Previously resolved · high confidence: The cooldown reset already functioned; explicit sequencing now spends the free Judgement promptly rather than resetting an almost-complete cooldown. The same-timestamp regression directly supports that implementation claim.

Current implementation / scenario: The reset already worked. Explicit APL placement now avoids resetting a nearly completed cooldown and casts the free Judgement before unrelated cooldowns.

Impact: About 31.5 to 35 Judgements per five minutes; roughly 5–6 DPS from the APL alone with shared support.

Remaining scope: Historical pooled uncertainty must respect overlapping seed ranges; no new timing-source gap was identified.

Implementation / model follow-up: Keep the sequencing fix and the SCEN-016 qualification on old comparison statistics.

sim/paladin/swift_judgement.go · ui/protection_paladin/apls/forever_protection.apl.json · tools/forever_bench/paladin_apl_test.go · artifacts/tank_support/validation.json

Related: PAL-014

Link to PAL-015

DONE-012 Removed and unavailable effects excluded Resolved · low

Question: Are old Sanctity, Molten Armor, Cat Mangle, campfire raid buffs or Shaman dual wield still granted?

Previously resolved · high confidence: Removed/unavailable choices are excluded from supported defaults and legality checks. Retained legacy records do not make Sanctity, Molten Armor, Cat Mangle, raid campfire effects or Shaman dual wield valid.

Current implementation / scenario: These unsupported choices are excluded; regular Faerie Fire works in supported forms. Surviving legacy records alone do not enable them.

Impact: Previous illegal/free-effect setups are not current baseline assumptions.

Remaining scope: A specific current acquisition source is required to reopen an availability exclusion.

Implementation / model follow-up: Keep exclusions explicit rather than reactivating a spell from its name or old ID.

docs/mechanics_review.md · docs/forever-70009.md · docs/spell_coverage.md · sim/core/equipment_validation.go

Earlier references: T36 / H12 / H25 / H27

Link to DONE-012

Complete readable register · Structured register

Method, gear and gain columns

All equipment is hypothetical level-65 modeled gear; real enchants remain. Hit comes from items, enchants, talents and racials, with no paid conversion.

Projection method and limits · Gear spreadsheet · Modeled item catalog · Gear-selection evidence · Stat/budget audit · Per-race gear/hit audit

Gear is selected by legal slot-coordinate comparisons, not an exhaustive combination search. Projected allocation budgets and special-stat prices are assumptions, not a recovered Blizzard formula. All selected items, enchants and scenario settings are retained in the replay profiles.

Gain columns average each available race’s percentage DPS change equally. Tier 1 compares bonuses on versus off with the same build. Gear scenarios increase item/suffix stats and weapon damage together; enchants, weapon speed/skill, procs, consumables and buffs stay fixed. Tier 1 stays on. Scaling amplification is the mean +50% gain divided by five times the mean +10% gain: 1× is linear, above 1× accelerates, below 1× flattens. A negative value means the +50% scenario loses DPS. An amplification dash means the +10% gain is too small or noisy for a useful ratio. This is finite-range curvature, not proof of exponential growth, isolated synergy or future item strength.

Replay files and downloads

Evidence and earlier reviews

These reports describe their pinned revisions. Current dispositions are in the register above.

DPS matrix

Sorted by each build’s peak mean DPS. Click a cell to load that exact race/build setup. The Ranked builds picker also loads complete profiles.

Hover cells for uncertainty and resource information. A dash means unavailable. * Separate build variants; their assumptions are in the register and build details.

Class / build	Alliance					Horde					Gains & scaling
Class / build	Human	Dwarf	Night Elf	Gnome	High Order	Orc	Tauren	Troll	Undead	Windshaper	Tier 1 gain	Gear +10%	Scaling amp.
Paladin Retribution	1005	1000	—	—	—	—	—	—	999	—	+1.7%	+5.1%	1.04×
Warrior Fury	990	987	980	982	1000	993	995	982	984	1000	+0.8%	+6.8%	1.04×
Hunter Beast Mastery	982	970	975	—	982	975	968	969	—	982	+2.4%	+4.4%	1.00×
Hunter Marksmanship	972	963	967	—	973	966	961	960	—	973	+5.9%	+5.5%	1.05×
Warrior Fury (Sunder)*	968	960	954	957	973	961	967	956	962	973	+0.8%	+6.9%	1.05×
Hunter Pet/Melee*	963	954	953	—	963	955	950	956	—	963	+0.4%	+5.1%	0.98×
Hunter Survival	959	952	953	—	956	955	951	947	—	956	-0.4%	+5.3%	1.01×
Paladin Physical Ret	950	946	—	—	—	—	—	—	936	—	+1.7%	+5.6%	1.04×
Warlock Demonic Pact	923	—	—	909	—	917	—	903	920	—	+1.6%	+4.2%	1.05×
Warrior Arms	870	868	865	858	868	869	869	859	863	868	+0.9%	+6.7%	1.04×
Mage Arcane–Frost*	857	—	—	852	856	851	—	839	850	—	+1.0%	+4.3%	1.03×
Mage Arcane	856	—	—	854	854	851	—	831	844	—	+1.4%	+4.3%	1.03×
Druid Feral	—	—	838	—	851	—	848	—	—	851	+2.7%	+4.4%	1.03×
Warrior 2H Bloodthirst*	843	841	836	834	844	846	842	837	839	844	+1.4%	+6.7%	1.02×
Warlock Affliction	844	—	—	822	—	839	—	823	831	—	+2.6%	+4.4%	1.05×
Priest Shadow	835	834	829	827	—	—	—	821	829	—	+0.3%	+3.4%	1.03×
Warlock Destruction	820	—	—	803	—	812	—	813	811	—	+2.2%	+4.1%	1.05×
Shaman Enhancement	—	816	—	—	—	817	815	807	—	818	+1.7%	+4.9%	0.98×
Warlock DS/Ruin	804	—	—	790	—	795	—	778	795	—	+2.6%	+4.9%	1.03×
Mage Fire	794	—	—	788	758	784	—	776	784	—	+1.2%	+4.8%	1.03×
Rogue Combat	762	762	758	758	774	760	—	755	760	774	+1.7%	+5.4%	1.00×
Rogue Mutilate	753	749	748	743	762	749	—	741	750	762	+2.3%	+4.7%	0.95×
Druid Balance	—	—	713	—	723	—	715	—	—	723	+2.0%	+5.0%	1.02×
Rogue Subtlety	706	704	703	705	718	705	—	701	704	718	+0.6%	+6.9%	0.83×
Mage Frost	643	—	—	653	635	639	—	628	634	—	+0.8%	+4.3%	0.97×
Druid Bear · Tank	—	—	545	—	551	—	552	—	—	551	+3.4%	+6.3%	1.04×
Shaman Elemental	—	532	—	—	—	534	531	526	—	526	+0.8%	+6.1%	1.06×
Paladin Protection · Tank	530	523	—	—	—	—	—	—	531	—	+1.6%	+4.0%	1.04×
Warrior Protection · Tank	508	507	497	507	510	502	514	505	517	510	+3.2%	+4.4%	0.96×
Shaman Stormcaller	—	469	—	—	—	469	473	462	—	461	+0.9%	+5.2%	1.06×
Priest Smite	451	447	446	456	—	—	—	442	457	—	+1.5%	+5.5%	1.07×

Tank performance

Tank rows take frontal attacks: 3,000 base damage every two seconds, parry haste, 1,500 HPS in three-second heals. Other DPS rows do not take these attacks. Tanks share raid support except their assigned self-maintained duties. Gear, talents and unverified class mechanics still differ; this is not a verified survival ranking.

Iterations continue after a modeled death. DPS is not discounted for death downtime; survival is evaluated separately.

Current correctness checks replay all 17 selected tank profiles in their original workloads; two disjoint 5,000-iteration ranges are pooled for the separate three-attacker table. October 11 corrects Revenge, Sunder and Thunder Clap threat/damage and Battle Shout threat, the October 11 log audit corrects Redoubt and Omen of Clarity, and the October 15 update (client 1.60.1.70425) changes racials, Reckoning, Righteous Fury, Consecration and Judgement coefficients, without reselecting tank profiles; Revenge uses a low-rank AP coefficient. 70291 selected 10 Warrior and 3 Paladin recipes; 4 Bear players are retained with corrected owned Faerie Fire. Central incoming Rage is UNVERIFIED. Six coefficient-20 sensitivity cases fail guards; small central survival tradeoffs and Multi DPS losses are explicitly reported. Point-estimate guard tolerances are not a 95% proof of no harm or global robustness. Current selections, sensitivity failures and limits. Shared support and Paladin oil/Judgement timing are included in the current replays; settings and evidence. Protection reserves rage and refreshes Sunder before stacks expire; stack coverage and validation.

Selection safeguards and three-attacker scenario · Matched validation and per-target threat · Tank metrics CSV · Research archives · Protection mana correction

Build / race	DPS	TPS	DTPS	TMI	Modeled death	3 attackers: minimum TPS	TPS by attacker
Bear · Tauren	552.0	1899.1	717.6	73.5	0.28%	376.8	1716, 377, 377
Bear · Windshaper	551.5	1888.9	715.7	78.4	0.48%	370.2	1691, 370, 370
Bear · High Order	551.5	1888.9	715.7	78.4	0.48%	370.2	1691, 370, 370
Bear · Night Elf	545.4	1876.3	700.3	77.9	0.50%	388.6	1742, 389, 389
Protection Paladin · Undead	531.2	898.4	704.0	77.2	0.16%	252.7	913, 262, 253
Protection Paladin · Human	530.2	916.7	702.4	78.4	0.18%	258.1	927, 268, 258
Protection Paladin · Dwarf	523.1	899.7	704.9	77.9	0.36%	255.8	907, 267, 256
Protection Warrior · Undead	516.6	1202.8	465.2	49.5	0.00%	60.9	1105, 181, 61
Protection Warrior · Tauren	513.7	1261.2	480.1	46.5	0.00%	60.7	1180, 202, 61
Protection Warrior · Windshaper	509.9	1197.6	486.0	49.9	0.00%	58.6	1121, 194, 59
Protection Warrior · High Order	509.9	1197.6	486.0	49.9	0.00%	58.6	1121, 194, 59
Protection Warrior · Human	508.2	1189.4	472.4	49.5	0.00%	59.0	1134, 206, 59
Protection Warrior · Dwarf	507.4	1268.8	496.8	50.6	0.00%	60.3	1174, 189, 60
Protection Warrior · Gnome	507.1	1194.0	482.0	49.5	0.00%	58.4	1117, 192, 58
Protection Warrior · Troll	504.8	1242.7	490.2	49.9	0.00%	59.9	1163, 198, 60
Protection Warrior · Orc	501.8	1256.7	504.6	51.4	0.00%	60.1	1159, 187, 60
Protection Warrior · Night Elf	497.2	1164.1	481.4	49.8	0.00%	57.7	1088, 189, 58