Discovery coordinator Root /root, session 01a12399-cfab-7c72-8455-e0e4cb7e156c, native branch feature/automated-combat-sweeps, native diagnostic #439. Read-only auditor /root/issue439_catalog_audit (GPT-6.1 Sol xhigh). Unassigned source investigation; no combat/rotation change is implemented by this finding.
Protection Paladin L60 cell cell-6e90f59f9c9a8fff9ef1 casts wow335-paladin-greater-blessing-of-sanctuary (rank Spell 25899, unlock 60) 73 times between 11.70 and 149.70 seconds, all targeting sweep:focal. Preserved raw events contain 73 actual completed-cast packets, 73 releases and 73 impacts, so this is not native collector inflation. The rank snapshot specifies 212 mana, 1,500 ms GCD and a 30-minute buff. There is no complete aura/dispel or cumulative mana-spend ledger, so maintained-state eligibility and exact spend still need reproduction.
The matched L50 run cell-71a6bfa0c8dbb1ddf616 lacks this spell. At L60, Consecration falls from 17 to two casts, 422 to seven damage packets, and 18,020 to 203 damage; 141 L60 resource refusals have 7–155 mana below its 393 cost. New Avenger's Shield has 71 resource refusals below 454 cost. Gear, mana capacity and individual Judgement damage increased. Both runs also retain long-range/LOS failures; the observed rate decline 155.73→53.51 damage/combat second is a maintenance/resource/access lead, not proof of lower damage coefficients.
Pinned source selector references:
src/game/wow335AbilityCatalog.ts:604–609 adds mitigation and 10% stamina modifiers.
src/game/benchmark/benchmarkController.ts:176–180 admits mitigation spells for tanks; candidates at 236–240.
Native RoM restrictions are filtered at benchmarkController 316–323, but the generic fallback loop at 360–397 can still attempt a candidate excluded by shared maintained-buff selection. Full earlier aura state is required to tie that path to each saved cast.
Acceptance for source investigation/fix:
Reproduce a learned active maintenance buff through the fallback path with explicit aura state and verify intended recast/refresh/expiry/dispel rules.
Prevent redundant buff casts if the maintained rule applies, while allowing legitimate expired/dispelled/other-target casts. Preserve source ability amounts, costs, GCDs and talent gates.
Add focused selector/source tests and record exact behavior and review before native adoption.
Keep route/LOS defects separate under native #19, and preserve the original once-only campaign rather than rerunning or rewriting it.
Exact frozen source snapshot: /Volumes/backup1tb-2/Projects/Issue-Fixes/native-rom-playable-party-2026-10-13/inputs/combat-source. Campaign/raw evidence: /Volumes/backup1tb-2/Projects/Issue-Fixes/automated-combat-sweeps-2026-10-14/campaign. Each source file is pinned in the frozen engine provenance; current source may have moved and must be checked before implementation.
Discovery coordinator Root `/root`, session `01a12399-cfab-7c72-8455-e0e4cb7e156c`, native branch `feature/automated-combat-sweeps`, [native diagnostic #439](https://git.whoagland.com/Codex/healer-man-godot/issues/439). Read-only auditor `/root/issue439_catalog_audit` (GPT-6.1 Sol xhigh). Unassigned source investigation; no combat/rotation change is implemented by this finding.
Protection Paladin L60 cell `cell-6e90f59f9c9a8fff9ef1` casts `wow335-paladin-greater-blessing-of-sanctuary` (rank Spell 25899, unlock 60) 73 times between 11.70 and 149.70 seconds, all targeting `sweep:focal`. Preserved raw events contain 73 actual completed-cast packets, 73 releases and 73 impacts, so this is not native collector inflation. The rank snapshot specifies 212 mana, 1,500 ms GCD and a 30-minute buff. There is no complete aura/dispel or cumulative mana-spend ledger, so maintained-state eligibility and exact spend still need reproduction.
The matched L50 run `cell-71a6bfa0c8dbb1ddf616` lacks this spell. At L60, Consecration falls from 17 to two casts, 422 to seven damage packets, and 18,020 to 203 damage; 141 L60 resource refusals have 7–155 mana below its 393 cost. New Avenger's Shield has 71 resource refusals below 454 cost. Gear, mana capacity and individual Judgement damage increased. Both runs also retain long-range/LOS failures; the observed rate decline 155.73→53.51 damage/combat second is a maintenance/resource/access lead, not proof of lower damage coefficients.
Pinned source selector references:
- `src/game/wow335AbilityCatalog.ts:604–609` adds mitigation and 10% stamina modifiers.
- `src/game/benchmark/benchmarkController.ts:176–180` admits mitigation spells for tanks; candidates at 236–240.
- Shared maintenance IDs/groups at 288–306 feed `evaluateCombatRotation`; `src/game/combatRotation.ts:416–419,624` recognizes/skips maintained buffs.
- Native RoM restrictions are filtered at benchmarkController 316–323, but the generic fallback loop at 360–397 can still attempt a candidate excluded by shared maintained-buff selection. Full earlier aura state is required to tie that path to each saved cast.
Acceptance for source investigation/fix:
- Reproduce a learned active maintenance buff through the fallback path with explicit aura state and verify intended recast/refresh/expiry/dispel rules.
- Prevent redundant buff casts if the maintained rule applies, while allowing legitimate expired/dispelled/other-target casts. Preserve source ability amounts, costs, GCDs and talent gates.
- Add focused selector/source tests and record exact behavior and review before native adoption.
- Keep route/LOS defects separate under native #19, and preserve the original once-only campaign rather than rerunning or rewriting it.
Exact frozen source snapshot: `/Volumes/backup1tb-2/Projects/Issue-Fixes/native-rom-playable-party-2026-10-13/inputs/combat-source`. Campaign/raw evidence: `/Volumes/backup1tb-2/Projects/Issue-Fixes/automated-combat-sweeps-2026-10-14/campaign`. Each source file is pinned in the frozen engine provenance; current source may have moved and must be checked before implementation.
Additional RoM evidence from #439's naturally saved once-only campaign, preserving source #18's unassigned scope: Root /root, session 01a12399-cfab-7c72-8455-e0e4cb7e156c; read-only auditor /root/issue439_catalog_audit (GPT-6.1 Sol xhigh).
Scout/Rogue L30 case cell-684d35be0f4a6d342e82 repeatedly uses Detection (rom-scout-primary-8, source skill490463, unlock26). Its snapshot specifies instant delivery, zero cooldown, 1,500 ms GCD,9focus cost and a300,000ms caster debuff reducing focus regeneration by5/3 per second. The preserved raw log has50 actual combat completed-cast events,50 releases and50 aura starts; these are three views of50 uses, not150 uses. Controller counters are73 calls,50 accepted and23 not-enough-focus refusals.
L20 lacks this skill. Observed rate falls45.22→33.90 damage/combat second while total damage rises4,244→4,370 and low-focus exposure increases. The runs are incomplete/source-limited, so this is a utility-maintenance/resource lead rather than a damage-coefficient conclusion.
Source selection mirrors the existing Paladin lead: src/game/benchmark/benchmarkController.ts:317–332 retains fallback candidates excluded by shared maintained-state selection; src/game/romFiveRotation.ts:270 guards hostile periodics/minimum hostiles but permits utilities by default; src/game/combatRotation.ts:624 filters maintained buffs/debuffs. Full aura/internal state is not archived, so exact candidate-path attribution remains qualified. Source issue #18 should include legitimate utility/debuff refresh/expiry rules, preserving intended casts while preventing repeated maintained effects if appropriate.
Raw source/campaign files are unchanged. Durable RoM triage under the central issue439 evidence will retain exact hashes and event proof. No new combat, source fix, coefficient change or native-readiness promotion occurred.
Additional RoM evidence from #439's naturally saved once-only campaign, preserving source #18's unassigned scope: Root `/root`, session `01a12399-cfab-7c72-8455-e0e4cb7e156c`; read-only auditor `/root/issue439_catalog_audit` (GPT-6.1 Sol xhigh).
Scout/Rogue L30 case `cell-684d35be0f4a6d342e82` repeatedly uses Detection (`rom-scout-primary-8`, source skill490463, unlock26). Its snapshot specifies instant delivery, zero cooldown, 1,500 ms GCD,9focus cost and a300,000ms caster debuff reducing focus regeneration by5/3 per second. The preserved raw log has50 actual combat completed-cast events,50 releases and50 aura starts; these are three views of50 uses, not150 uses. Controller counters are73 calls,50 accepted and23 not-enough-focus refusals.
L20 lacks this skill. Observed rate falls45.22→33.90 damage/combat second while total damage rises4,244→4,370 and low-focus exposure increases. The runs are incomplete/source-limited, so this is a utility-maintenance/resource lead rather than a damage-coefficient conclusion.
Source selection mirrors the existing Paladin lead: `src/game/benchmark/benchmarkController.ts:317–332` retains fallback candidates excluded by shared maintained-state selection; `src/game/romFiveRotation.ts:270` guards hostile periodics/minimum hostiles but permits utilities by default; `src/game/combatRotation.ts:624` filters maintained buffs/debuffs. Full aura/internal state is not archived, so exact candidate-path attribution remains qualified. Source issue #18 should include legitimate utility/debuff refresh/expiry rules, preserving intended casts while preventing repeated maintained effects if appropriate.
Raw source/campaign files are unchanged. Durable RoM triage under the central issue439 evidence will retain exact hashes and event proof. No new combat, source fix, coefficient change or native-readiness promotion occurred.
Additional confirmed utility-recast evidence from the completed #439 frozen once-only campaign. Blood Death Knight L80 cell-2615c541b65c4f106081 completes Dark Command spell56222 3,509 times: source diagnostics3,591calls/3,509accepted/82rejected; raw3,509 combat completions,3,509 releases and3,509 impacts. Initial uses include0.05/0.10/0.15 seconds on the same target. These are actual repeated uses, not lifecycle inflation or just rejected attempts. Matched L70 cell-6939308d61782776b873 has3accepted of295calls. Native source reviewer /root/issue439_simulator (GPT-6.1 Sol xhigh) checked raw events and frozen source hashes. Exact selector/aggro/cooldown eligibility remains an investigation rather than an asserted full-state reproduction.
The separate Death Pact hostile execute/Blood Gorged active attack conversion problem is now shared #19#19; keep that output inflation distinct from this repeated utility-selection lead. No source, engine or combat measurement changes were made. Root /root session01a12399-cfab-7c72-8455-e0e4cb7e156c. Evidence /Volumes/backup1tb-2/Projects/Issue-Fixes/automated-combat-sweeps-2026-10-14/campaign/cells and final reviewer receipt.
Additional confirmed utility-recast evidence from the completed #439 frozen once-only campaign. Blood Death Knight L80 cell-2615c541b65c4f106081 completes Dark Command spell56222 3,509 times: source diagnostics3,591calls/3,509accepted/82rejected; raw3,509 combat completions,3,509 releases and3,509 impacts. Initial uses include0.05/0.10/0.15 seconds on the same target. These are actual repeated uses, not lifecycle inflation or just rejected attempts. Matched L70 cell-6939308d61782776b873 has3accepted of295calls. Native source reviewer /root/issue439_simulator (GPT-6.1 Sol xhigh) checked raw events and frozen source hashes. Exact selector/aggro/cooldown eligibility remains an investigation rather than an asserted full-state reproduction.
The separate Death Pact hostile execute/Blood Gorged active attack conversion problem is now shared #19 https://git.whoagland.com/phenom/healer-man/issues/19; keep that output inflation distinct from this repeated utility-selection lead. No source, engine or combat measurement changes were made. Root /root session01a12399-cfab-7c72-8455-e0e4cb7e156c. Evidence /Volumes/backup1tb-2/Projects/Issue-Fixes/automated-combat-sweeps-2026-10-14/campaign/cells and final reviewer receipt.
Follow-up to Dark Command comment7818: pure frozen-source function evaluation now identifies Morbidity rank3 named Death-and-Decay -15000ms cooldown scope incorrectly applying to DarkCommand8000ms, clamping to0. This precise parser/applicability defect is tracked separately at #20. Keep Sanctuary/Detection maintenance findings open here; DarkCommand is no longer attributed simply to redundant maintained-effect fallback. No source/fight changes. Root /root session01a12399-cfab-7c72-8455-e0e4cb7e156c; GPT-6.1 Sol xhigh source evaluator /root/issue439_simulator.
Follow-up to Dark Command comment7818: pure frozen-source function evaluation now identifies Morbidity rank3 named Death-and-Decay -15000ms cooldown scope incorrectly applying to DarkCommand8000ms, clamping to0. This precise parser/applicability defect is tracked separately at https://git.whoagland.com/phenom/healer-man/issues/20. Keep Sanctuary/Detection maintenance findings open here; DarkCommand is no longer attributed simply to redundant maintained-effect fallback. No source/fight changes. Root /root session01a12399-cfab-7c72-8455-e0e4cb7e156c; GPT-6.1 Sol xhigh source evaluator /root/issue439_simulator.
Owner-authorized fix now claimed. Coordinator Root /root, session 01a12399-cfab-7c72-8455-e0e4cb7e156c. Worker /root/combat_rotation_luna, model/reasoning gpt-6-luna / xhigh, branch fix/18-maintenance-cast-selection, source base a9322a3897. Scope: Reproduce and repair maintenance eligibility through automatic fallback; selector/rotation files, no combatStore edits. Isolated source worktrees preserve all native dirty work and frozen #439 archives. Focused source tests and Ready-five checks, independent review and source publication precede separately pinned native composition/adoption. No all-class campaign is repeated. The owner requested GPT-6 Luna xhigh workers for the simpler tasks.
Owner-authorized fix now claimed. Coordinator Root /root, session 01a12399-cfab-7c72-8455-e0e4cb7e156c. Worker /root/combat_rotation_luna, model/reasoning gpt-6-luna / xhigh, branch fix/18-maintenance-cast-selection, source base a9322a389709df5fb87c2423a825f09580a2d8fe. Scope: Reproduce and repair maintenance eligibility through automatic fallback; selector/rotation files, no combatStore edits. Isolated source worktrees preserve all native dirty work and frozen #439 archives. Focused source tests and Ready-five checks, independent review and source publication precede separately pinned native composition/adoption. No all-class campaign is repeated. The owner requested GPT-6 Luna xhigh workers for the simpler tasks.
Source implementation independently reviewed and published in #21 at exact head 2f8cb8a510. Combined validation: 94 focused tests, Ready-five 7 JavaScript +20 Python, full source TypeScript and required release check passed. Root /root session01a12399-cfab-7c72-8455-e0e4cb7e156c coordinates the named Luna xhigh workers. Native source composition/engine adoption is tracked separately by Codex/healer-man-godot#457; historical 873-cell evidence remains unchanged. Memory and final source/main/adoption states will be recorded before acceptance closure.
Source implementation independently reviewed and published in https://git.whoagland.com/phenom/healer-man/pulls/21 at exact head 2f8cb8a5104978e66077523202b4d7f9ec24ff60. Combined validation: 94 focused tests, Ready-five 7 JavaScript +20 Python, full source TypeScript and required release check passed. Root /root session01a12399-cfab-7c72-8455-e0e4cb7e156c coordinates the named Luna xhigh workers. Native source composition/engine adoption is tracked separately by Codex/healer-man-godot#457; historical 873-cell evidence remains unchanged. Memory and final source/main/adoption states will be recorded before acceptance closure.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Discovery coordinator Root
/root, session01a12399-cfab-7c72-8455-e0e4cb7e156c, native branchfeature/automated-combat-sweeps, native diagnostic #439. Read-only auditor/root/issue439_catalog_audit(GPT-6.1 Sol xhigh). Unassigned source investigation; no combat/rotation change is implemented by this finding.Protection Paladin L60 cell
cell-6e90f59f9c9a8fff9ef1castswow335-paladin-greater-blessing-of-sanctuary(rank Spell 25899, unlock 60) 73 times between 11.70 and 149.70 seconds, all targetingsweep:focal. Preserved raw events contain 73 actual completed-cast packets, 73 releases and 73 impacts, so this is not native collector inflation. The rank snapshot specifies 212 mana, 1,500 ms GCD and a 30-minute buff. There is no complete aura/dispel or cumulative mana-spend ledger, so maintained-state eligibility and exact spend still need reproduction.The matched L50 run
cell-71a6bfa0c8dbb1ddf616lacks this spell. At L60, Consecration falls from 17 to two casts, 422 to seven damage packets, and 18,020 to 203 damage; 141 L60 resource refusals have 7–155 mana below its 393 cost. New Avenger's Shield has 71 resource refusals below 454 cost. Gear, mana capacity and individual Judgement damage increased. Both runs also retain long-range/LOS failures; the observed rate decline 155.73→53.51 damage/combat second is a maintenance/resource/access lead, not proof of lower damage coefficients.Pinned source selector references:
src/game/wow335AbilityCatalog.ts:604–609adds mitigation and 10% stamina modifiers.src/game/benchmark/benchmarkController.ts:176–180admits mitigation spells for tanks; candidates at 236–240.evaluateCombatRotation;src/game/combatRotation.ts:416–419,624recognizes/skips maintained buffs.Acceptance for source investigation/fix:
Exact frozen source snapshot:
/Volumes/backup1tb-2/Projects/Issue-Fixes/native-rom-playable-party-2026-10-13/inputs/combat-source. Campaign/raw evidence:/Volumes/backup1tb-2/Projects/Issue-Fixes/automated-combat-sweeps-2026-10-14/campaign. Each source file is pinned in the frozen engine provenance; current source may have moved and must be checked before implementation.Additional RoM evidence from #439's naturally saved once-only campaign, preserving source #18's unassigned scope: Root
/root, session01a12399-cfab-7c72-8455-e0e4cb7e156c; read-only auditor/root/issue439_catalog_audit(GPT-6.1 Sol xhigh).Scout/Rogue L30 case
cell-684d35be0f4a6d342e82repeatedly uses Detection (rom-scout-primary-8, source skill490463, unlock26). Its snapshot specifies instant delivery, zero cooldown, 1,500 ms GCD,9focus cost and a300,000ms caster debuff reducing focus regeneration by5/3 per second. The preserved raw log has50 actual combat completed-cast events,50 releases and50 aura starts; these are three views of50 uses, not150 uses. Controller counters are73 calls,50 accepted and23 not-enough-focus refusals.L20 lacks this skill. Observed rate falls45.22→33.90 damage/combat second while total damage rises4,244→4,370 and low-focus exposure increases. The runs are incomplete/source-limited, so this is a utility-maintenance/resource lead rather than a damage-coefficient conclusion.
Source selection mirrors the existing Paladin lead:
src/game/benchmark/benchmarkController.ts:317–332retains fallback candidates excluded by shared maintained-state selection;src/game/romFiveRotation.ts:270guards hostile periodics/minimum hostiles but permits utilities by default;src/game/combatRotation.ts:624filters maintained buffs/debuffs. Full aura/internal state is not archived, so exact candidate-path attribution remains qualified. Source issue #18 should include legitimate utility/debuff refresh/expiry rules, preserving intended casts while preventing repeated maintained effects if appropriate.Raw source/campaign files are unchanged. Durable RoM triage under the central issue439 evidence will retain exact hashes and event proof. No new combat, source fix, coefficient change or native-readiness promotion occurred.
Additional confirmed utility-recast evidence from the completed #439 frozen once-only campaign. Blood Death Knight L80 cell-2615c541b65c4f106081 completes Dark Command spell56222 3,509 times: source diagnostics3,591calls/3,509accepted/82rejected; raw3,509 combat completions,3,509 releases and3,509 impacts. Initial uses include0.05/0.10/0.15 seconds on the same target. These are actual repeated uses, not lifecycle inflation or just rejected attempts. Matched L70 cell-6939308d61782776b873 has3accepted of295calls. Native source reviewer /root/issue439_simulator (GPT-6.1 Sol xhigh) checked raw events and frozen source hashes. Exact selector/aggro/cooldown eligibility remains an investigation rather than an asserted full-state reproduction.
The separate Death Pact hostile execute/Blood Gorged active attack conversion problem is now shared #19 #19; keep that output inflation distinct from this repeated utility-selection lead. No source, engine or combat measurement changes were made. Root /root session01a12399-cfab-7c72-8455-e0e4cb7e156c. Evidence /Volumes/backup1tb-2/Projects/Issue-Fixes/automated-combat-sweeps-2026-10-14/campaign/cells and final reviewer receipt.
Follow-up to Dark Command comment7818: pure frozen-source function evaluation now identifies Morbidity rank3 named Death-and-Decay -15000ms cooldown scope incorrectly applying to DarkCommand8000ms, clamping to0. This precise parser/applicability defect is tracked separately at #20. Keep Sanctuary/Detection maintenance findings open here; DarkCommand is no longer attributed simply to redundant maintained-effect fallback. No source/fight changes. Root /root session01a12399-cfab-7c72-8455-e0e4cb7e156c; GPT-6.1 Sol xhigh source evaluator /root/issue439_simulator.
Owner-authorized fix now claimed. Coordinator Root /root, session 01a12399-cfab-7c72-8455-e0e4cb7e156c. Worker /root/combat_rotation_luna, model/reasoning gpt-6-luna / xhigh, branch fix/18-maintenance-cast-selection, source base
a9322a3897. Scope: Reproduce and repair maintenance eligibility through automatic fallback; selector/rotation files, no combatStore edits. Isolated source worktrees preserve all native dirty work and frozen #439 archives. Focused source tests and Ready-five checks, independent review and source publication precede separately pinned native composition/adoption. No all-class campaign is repeated. The owner requested GPT-6 Luna xhigh workers for the simpler tasks.Source implementation independently reviewed and published in #21 at exact head
2f8cb8a510. Combined validation: 94 focused tests, Ready-five 7 JavaScript +20 Python, full source TypeScript and required release check passed. Root /root session01a12399-cfab-7c72-8455-e0e4cb7e156c coordinates the named Luna xhigh workers. Native source composition/engine adoption is tracked separately by Codex/healer-man-godot#457; historical 873-cell evidence remains unchanged. Memory and final source/main/adoption states will be recorded before acceptance closure.