Problem
There is no record of what a launch actually consisted of. When a launch misbehaves — wrong content, wrong variant, unexpected engine build — there is nothing to compare against.
Proposed approach
Record a receipt per launch capturing:
- Retail archive roots
- Archive fingerprints
- Environment
- Variant
- Engine identity
Then perform cheap revalidation against the receipt before a subsequent launch, so drift is detected without a full re-scan.
Why it matters beyond debugging
This is a prerequisite for LAN and determinism work: without a verifiable record of what each participant launched, desync investigation is guesswork.
Acceptance criteria
- Every launch produces a receipt.
- Pre-launch revalidation is cheap enough to run every time.
- Drift between receipt and current state is detected and surfaced.
Problem
There is no record of what a launch actually consisted of. When a launch misbehaves — wrong content, wrong variant, unexpected engine build — there is nothing to compare against.
Proposed approach
Record a receipt per launch capturing:
Then perform cheap revalidation against the receipt before a subsequent launch, so drift is detected without a full re-scan.
Why it matters beyond debugging
This is a prerequisite for LAN and determinism work: without a verifiable record of what each participant launched, desync investigation is guesswork.
Acceptance criteria