# Scene Skill 102 Pseudo-Production Batch Execution Plan

> Date: 2026-04-20
> Status: Draft
> Upstream Design: `docs/superpowers/specs/2026-04-20-scene-skill-102-pseudoprod-batch-execution-design.md`

## Plan Intent

Run the prepared 10-scene pseudo-production batch in an operator-provided environment and record structured results.

This plan is bounded to execution and evidence collection for the selected 10 scenes.

## Fixed Inputs

1. `tests/fixtures/generated_scene/scene_skill_102_pseudoprod_execution_handoff_2026-04-20.json`
2. `tests/fixtures/generated_scene/scene_skill_102_pseudoprod_evidence_checklist_2026-04-20.json`
3. `tests/fixtures/generated_scene/scene_skill_102_pseudoprod_execution_record_template_2026-04-20.json`
4. `tests/fixtures/generated_scene/scene_skill_102_pseudoprod_batch_selection_2026-04-20.json`

## Planned Outputs

1. `tests/fixtures/generated_scene/scene_skill_102_pseudoprod_batch_execution_results_2026-04-20.json`
2. `docs/superpowers/reports/2026-04-20-scene-skill-102-pseudoprod-batch-execution-report.md`

## Allowed Files

1. planned execution result JSON
2. planned execution report
3. redacted evidence summaries if explicitly generated

## Forbidden Files

1. `src/generated_scene/analyzer.rs`
2. `src/generated_scene/generator.rs`
3. `src/generated_scene/ir.rs`
4. `examples/scene_skill_102_final_materialization_2026-04-19/skills/**`
5. `tests/fixtures/generated_scene/scene_execution_board_2026-04-18.json`
6. any credential, token, cookie, or secret file

## Phase 0: Confirm Environment Readiness

### Tasks

1. Confirm operator-provided browser/runtime environment exists.
2. Confirm network/session access is provided outside repository.
3. Confirm evidence output location.
4. Confirm redaction rules.

### Acceptance Criteria

1. No credentials are stored in repository.
2. Execution does not start unless environment readiness is confirmed externally.

## Phase 1: Execute Selected Scenes

### Tasks

For each selected scene:

1. use the deterministic invocation input ending with `。。。`
2. execute through sgClaw runtime or agreed quasi-production host
3. collect console log
4. collect network summary
5. capture screenshot if target page is required
6. capture exported artifact if produced
7. record final result state

### Acceptance Criteria

1. every selected scene has one execution record
2. every record has exactly one result state
3. failures use the allowed taxonomy

## Phase 2: Redact And Normalize Evidence

### Tasks

1. redact credentials, cookies, tokens, Authorization headers, and private data
2. normalize evidence paths
3. confirm each evidence checklist item is present or explicitly unavailable

### Acceptance Criteria

1. no secret material enters repository output
2. missing evidence has a reason

## Phase 3: Publish Execution Results

### Tasks

1. write execution results JSON
2. write execution report
3. summarize pass/blocker/mismatch/runtime-error counts
4. list follow-up blockers

### Acceptance Criteria

1. selected scene count remains 10
2. report does not claim full production certification
3. official board is not updated under this plan

## Completion Criteria

This plan is complete when all 10 selected scenes have structured execution records and a redacted execution report is published.

## Stop Statement

Stop after publishing execution results and report.

Do not update official board status under this plan.