Files
aetherbound-guild/docs/runtime/PHASE9_0R7_INDIVIDUAL_GROWTH_CONTRACT.md

4.5 KiB

Phase 9.0R7 Individual Growth Runtime Contract

Authority And Boundary

This contract owns only the continuation from the accepted P9.0R6 second-cycle result through one individual level choice, one fifth battle and one third-cycle result. It previews the product authority that levels belong to individual generated Recruits, but does not implement the authoritative XP, promotion, Market, save or content systems.

Stable State

member_slots = [front, rear] + [added when present]
member_level(slot) = 2 only when slot == trained_member; otherwise 1
veteran_mark_available = route_reward_committed and not training_committed
training_choice = empty | one current member slot
training_committed = false -> true exactly once
fifth_reward_committed = false -> true exactly once

Slots bind to the current runtime member identity and profession. An absent added slot never renders and is illegal at selection, review and commit.

Training Transaction

  • open_training requires the R6 result and committed route reward, clears only the uncommitted training selection and opens the training scene.
  • choose_training accepts one current member slot and mutates selection only.
  • open_training_review requires one legal selection and mutates nothing.
  • cancel_training_review returns to the same selected training scene.
  • commit_training revalidates the selected slot, consumes the mark, makes exactly that slot Lv.2, locks the transaction and starts the fifth battle.
  • Duplicate or out-of-page commit is inert.

Fifth Battle

The fifth encounter uses existing accepted actors and Riftblade-family assets. It is a bounded presentation fixture, not an authoritative combat engine or a new catalog identity.

Base state:

front_hp = 100
rear_hp = 100
enemy_hp = 140
base_time = 7.40s
Farstring Lv.2 time = 6.40s
reward = +8 Coin at verified result

Ordered consequences:

  1. A trained Farstring moves its one normal shot before enemy pressure; it does not gain a second attack.
  2. Front crush deals -45; retained EQ-002 reduces it to -30; trained front reduces the already-derived damage by another 20.
  3. Split pressure hits the original rear for -40.
  4. Original rear acts once here unless its trained Farstring shot already moved to step 1.
  5. Added rear acts once here when present unless its trained Farstring shot already moved to step 1; otherwise the retained build follows up.
  6. Route guard reaches 0 and reward commits once.

Kindhand healing targets the damaged original rear. A normal Kindhand restores +12 to rear HP 72; a trained Kindhand restores +25 to rear HP 85. Farstring training changes timing only. Front training changes front HP only. Both rear professions act once in a full line.

Result Matrix

Trained role Frontpin retained Front HP Rear HP with Kindhand Time
front yes 90 72 7.40
front no 75 72 7.40
Farstring yes 70 72 when Kindhand present 6.40
Farstring no 55 72 when Kindhand present 6.40
Kindhand yes 70 85 7.40
Kindhand no 55 85 7.40

When a two-person Company has no Kindhand, rear HP remains 60. When Kindhand is present but not trained, rear HP becomes 72. Route selection changes the inherited Coin only; it does not secretly change this fixed fifth encounter.

Presentation

  • Training is a scene with the actual Company standing in the world and one selectable level crest above or beside each member, not a generic settings list or spreadsheet.
  • Every member shows position, profession, current level and the exact next event change before selection.
  • Review names the member and single counterfactual, then offers Back and Train.
  • Fifth battle visibly distinguishes early shot, reduced front damage and stronger rear recovery using existing number/VFX feedback.
  • Result names the causal level effect before Coin/HP/time totals.
  • Primary 844x390 has no scrolling and every action is at least 44px.

Evidence

  • Focused tests cover legal member sets, training transaction, member levels, all role outcomes, exact times, reward and cleanup.
  • Real parsed ScreenTouch covers open, revise, review, cancel, commit, battle and result with no fallback or direct callback path.
  • Normal-render captures cover two- and three-person training, review, the three role effects and representative results.
  • Fast preview evidence cannot become formal candidate, human, device or release evidence.