aetherbound-guild/docs/runtime/PHASE9_0R6_SECOND_CYCLE_ROUTE_CONTRACT.md

Phase 9.0R6 Second-Cycle Route Runtime Contract

State Machine

P9.0R5 session_result
  -> route_board_unselected
  -> route_board_selected
  -> route_commit_review
  -> fourth_battle
  -> second_cycle_result

Legal reverse paths are selected route to another route and commit review back to the same selected route. Route commit has no reverse path. Home and confirmed Restart clear the complete P9.0R1-R6 transient experience.

Stable Route Offers

ID	Name	Pressure	Exposure	Reward
R6-01	Windwall Causeway / 风墙长堤	captain front crush	low	+4 Coin
R6-02	Bellglass Run / 钟晶疾道	early tempo burst	medium	+7 Coin
R6-03	Twinflare Ravine / 双辉裂谷	split rear strike	high	+10 Coin

The route board states known pressure, exposure, reward and current coverage. It does not expose exact future HP, mark an option recommended or preselect a route. Route identities are independently authored for this game.

Coverage Rules

build_profile =
  guarded   when retained_item == EQ-002
  tempo     when retained_item == EQ-061
  full_line when added_rear is Farstring or Kindhand

Route	guarded	tempo	full_line
R6-01	covered: retained brace answers front crush	gap: initiative does not reduce the hit	partial: Kindhand can recover after the hit
R6-02	gap: brace does not answer enemy tempo	covered: Early Bell acts before the burst	partial: broader rear line recovers but does not interrupt
R6-03	gap: front brace does not protect split rear	gap: initiative does not protect split rear	covered: Farstring and Kindhand divide damage and recovery

Route Transaction

• open_route_board inherits exact current Coin, Company and build state.
• choose_route validates one stable ID and mutates selection only.
• review_route requires one selection and mutates nothing.
• cancel_review returns to the selected board state.
• commit_route is the only route commit, locks the current snapshot once and starts the fourth battle. Duplicate commit is inert.
• Route entry costs 0 Coin; reward is not granted until battle result.

Fourth Battle Matrix

Route/build	Duration	Front after pressure	Rear after pressure/support
R6-01 guarded	7.40	90, then 100 only if original rear is Kindhand	100
R6-01 tempo	6.40	60, then 72 only if original rear is Kindhand	100
R6-01 full_line	7.00	70 -> 82	100
R6-02 guarded	8.00	70, then 82 only if original rear is Kindhand	100
R6-02 tempo	6.00	70, then 82 only if original rear is Kindhand	100
R6-02 full_line	6.80	70 -> 82	100
R6-03 guarded	7.80	100	100 -> 65; Kindhand self-supports to 77
R6-03 tempo	6.70	100	100 -> 65; Kindhand self-supports to 77
R6-03 full_line	7.00	100	100 -> 65 -> 88

Every HP mutation emits one existing P9.0R3 feedback payload over the actual front, original rear or enemy actor. Both rear professions act once on the full-line route. Enemy reaches 0; no respawn, casualty, replacement or save state is created.

Reward And Result

Reward commit is exactly once after enemy defeat:

R6-01: coin += 4
R6-02: coin += 7
R6-03: coin += 10

Result order is build profile and route, pressure/coverage consequence, exact HP/time, one-time reward and Coin/Company, then Restart/Home only.

Presentation

• Route selection is a world/map scene with three adjacent route nodes and the current Company visible, not a settings form or receipt table.
• Primary 844x390 viewport has no scrolling and every action is at least 44px.
• Front crush uses a front danger cue; tempo uses an early action cue; rear split marks the actual rear actor and places its number feedback there.
• Existing art is a bounded Riftblade-family presentation; no new enemy catalog identity is claimed.

Fast Evidence

• Focused test covers 3x3 coverage, all matrix outcomes, original-rear counterfactuals, selection/revision, review cancel, once-only commit/reward, invalid input and cleanup.
• Parsed ScreenTouch continues from a real P9.0R5 session result through route selection, review, cancel/revise, battle and result with fallback/direct 0.
• Normal-render captures cover route unselected, each build coverage, review, the three pressure languages and representative results.
• P9.0R5/P9.0R4/P9.0R3 focused regressions remain mandatory.
• Fast preview evidence cannot become formal candidate evidence.