claw/docs/superpowers/plans/2026-04-19-g7-real-sample-entry-plan.md

G7 Real-Sample Entry Plan

Date: 2026-04-19 Status: Draft Upstream Spec: 2026-04-19-g7-real-sample-entry-design.md

Plan Intent

This plan executes one bounded next slice:

G7 real-sample entry

Scope Guardrails

1. do not reopen mainline families
2. do not execute G6 or G8
3. do not add new G7 family fixtures
4. do not implement new runtime-platform prerequisites under this plan
5. do not open G4 / G5

Fixed Verification Anchor

The only target under this plan is:

1. 计量资产库存统计

Workstreams

1. WS1 Freeze the G7 real-sample boundary
2. WS2 Build the real-sample contract differential
3. WS3 Rerun the fixed real sample against the existing G7 runtime contract
4. WS4 Update validation assets and close the loop

Phase 0: Freeze the Boundary

Objective

Lock the plan to one G7 representative sample.

Tasks

1. freeze 计量资产库存统计 as the only real-sample anchor
2. freeze existing G7 repo-local runtime contract as the starting baseline

Acceptance Criteria

1. no other boundary family is touched under this plan

Phase 1: Build the Differential

Objective

Understand whether the existing G7 runtime contract is already close enough for a real-sample rerun.

Tasks

1. compare the representative G7 fixture contract to the chosen real sample
2. isolate the smallest remaining contract risk

Acceptance Criteria

1. the rerun target is explicit and bounded

Phase 2: Real-Sample Rerun

Objective

Use the fixed real sample to test the current G7 runtime contract.

Tasks

1. run sg_scene_generate on the fixed G7 real sample
2. record whether the result is a pass or a smaller mismatch

Acceptance Criteria

1. the outcome is narrower than not yet executed

Phase 3: Validation Closure

Objective

Write the result back into the validation layer and close the bounded slice.

Tasks

1. update validation assets if the outcome narrows
2. write a closure report

Acceptance Criteria

1. the next boundary-family ambiguity is reduced further without broadening roadmap scope