# Remaining Route Conflict Correction Plan

> Date: 2026-04-19
> Status: Draft
> Upstream Spec: `docs/superpowers/specs/2026-04-19-remaining-route-conflict-correction-design.md`

## Plan Intent

Adjudicate and, where evidence supports it, correct the remaining `4` route conflicts from the follow-up full sweep.

This is a bounded route-conflict plan, not a new full-sweep roadmap.

## Fixed Input

Use only the `4` `misclassified` records from:

`tests/fixtures/generated_scene/full_sweep_improvement_followup_2026-04-19.json`

The fixed scene set is:

1. `95598报修工单日管控`
2. `95598重要服务事项报备统计表`
3. `台区线损台区月度高负损预测`
4. `配网支撑月报(95598抢修统计报表)`

## Fixed Outputs

1. `tests/fixtures/generated_scene/remaining_route_conflict_decisions_2026-04-19.json`
2. `docs/superpowers/reports/2026-04-19-remaining-route-conflict-correction-report.md`

## Scope Guardrails

1. do not touch timeout handling
2. do not touch structured fail-closed reporting
3. do not add new families
4. do not update execution board
5. do not promote scenes
6. do not weaken current `G2/G3/G6` pass cases
7. do not force a scene into G2/G3 if host bridge is the only complete path

## Phase 0: Freeze Conflict Set

### Objective

Freeze the `4` route conflicts as the only input.

### Tasks

1. read `full_sweep_improvement_followup_2026-04-19.json`
2. select only `dryRunStatus = misclassified`
3. verify count is `4`
4. freeze expected group and inferred archetype for each record

### Deliverables

1. frozen route conflict inventory

### Acceptance Criteria

1. exactly `4` records are in scope
2. no extra scene is added

## Phase 1: Evidence Adjudication

### Objective

Decide whether each conflict should be corrected or retained as host bridge.

### Tasks

1. inspect existing generation reports for the `4` records
2. compare business-chain evidence against host-bridge evidence
3. apply the route decision model:
   - `route-corrected-to-g3`
   - `route-corrected-to-g2`
   - `valid-host-bridge-workflow`
   - `board-expectation-stale`
   - `route-conflict-unresolved`
4. write preliminary decisions

### Deliverables

1. preliminary route conflict decision table

### Acceptance Criteria

1. all `4` records have a preliminary decision
2. no code is changed before evidence is adjudicated

## Phase 2: Bounded Route Correction

### Objective

Apply only the route corrections justified by Phase 1.

### Tasks

1. update analyzer routing precedence only if evidence supports correction
2. keep valid host-bridge cases unchanged
3. add targeted regression tests for corrected cases
4. preserve existing `G2/G3/G6` real-sample and canonical tests

### Deliverables

1. bounded analyzer routing patch if needed
2. route conflict regression tests

### Acceptance Criteria

1. corrected records no longer misclassify
2. valid host-bridge records remain host bridge
3. no broad routing rewrite is introduced

## Phase 3: Targeted Probe

### Objective

Verify only the fixed `4` records after correction.

### Tasks

1. rerun generation for the same `4` scenes
2. record resulting archetype and readiness
3. classify each final decision
4. write final decision JSON

### Deliverables

1. `remaining_route_conflict_decisions_2026-04-19.json`

### Acceptance Criteria

1. all `4` records have final probe results
2. no full `102` sweep is required by this plan

## Phase 4: Report and Stop

### Objective

Publish the route conflict report and stop.

### Tasks

1. write the route conflict correction report
2. include final decisions for all `4` records
3. list verification commands
4. explicitly state that the execution board is not updated

### Deliverables

1. route conflict correction report

### Acceptance Criteria

1. all `4` conflicts are adjudicated
2. tests pass
3. no execution board update is made

## Completion Criteria

This plan is complete when:

1. the fixed `4` route conflicts have final decisions
2. targeted probes have been run
3. relevant regressions pass
4. decision JSON and report are published
5. execution stops without opening another plan