| # Methodology | |
| ## Goal | |
| Produce the strongest practical modernization proof possible without paid or licensed legacy runtime infrastructure. | |
| ## Method | |
| 1. Pin the upstream AWS CardDemo source repository and commit. | |
| 2. Hash the source files and generated modern files. | |
| 3. Extract capabilities from legacy source, copybooks, BMS maps, JCL, seed data, and static contracts. | |
| 4. Build a one-for-one capability ledger. | |
| 5. Implement the modern Python/SQLite target against those source-derived contracts. | |
| 6. Generate receipts, golden traces, edge cases, and parity reports. | |
| 7. Execute the modern service against static/source-oracle replay checks. | |
| 8. Run clean-copy reproduction and API smoke checks. | |
| 9. Run a strict A-grade gate that refuses promotion without live legacy runtime traces. | |
| 10. Publish the evidence with explicit boundaries. | |
| ## Current Proof State | |
| - B-grade static/source-oracle capabilities: `50` | |
| - F-grade gaps: `0` | |
| - Gap count: `0` | |
| - Static replay passed: `True` | |
| - No-cost confidence posture: `strong_pre_runtime_confidence` | |
| - A-grade promotable: `False` | |
| ## What Would Upgrade This To A | |
| The A-grade gate requires accepted live legacy runtime trace records for every capability. Each record must bind a capability ID to the pinned source head, a live runtime trace kind, a raw legacy trace hash, a modern observation hash, a comparator version, and `matched=true`. | |
| The current bundle has zero accepted live runtime matches because no paid/licensed legacy runtime was executed. | |