|
| 1 | +{ |
| 2 | + "schema_version": "1.0", |
| 3 | + "required_cases": [ |
| 4 | + "single_turn_text", |
| 5 | + "multi_turn_text", |
| 6 | + "tool_call_response", |
| 7 | + "state_shallow_update", |
| 8 | + "memory_scope_user_session", |
| 9 | + "summary_create_update", |
| 10 | + "summary_event_truncation", |
| 11 | + "failure_retry_append", |
| 12 | + "cross_session_isolation", |
| 13 | + "summary_defect_specials" |
| 14 | + ], |
| 15 | + "extended_cases": [ |
| 16 | + "all_entities_contract" |
| 17 | + ], |
| 18 | + "requirements": { |
| 19 | + "p0_backends": { |
| 20 | + "requirement": "Run lightweight replay against InMemory and a persistent or simulated persistent backend.", |
| 21 | + "backends": ["in_memory", "sqlite"], |
| 22 | + "test": "test_lightweight_replay_cases" |
| 23 | + }, |
| 24 | + "p0_public_cases": { |
| 25 | + "requirement": "Provide at least 10 deterministic public replay cases.", |
| 26 | + "required_case_count": 10, |
| 27 | + "extended_case_count": 1, |
| 28 | + "test": "test_acceptance_matrix_matches_fixtures" |
| 29 | + }, |
| 30 | + "p0_public_case_mutation_matrix": { |
| 31 | + "requirement": "Each required public replay case detects an injected inconsistency.", |
| 32 | + "detected": 10, |
| 33 | + "total": 10, |
| 34 | + "rate": 1.0, |
| 35 | + "test": "test_required_public_cases_detect_injected_inconsistency" |
| 36 | + }, |
| 37 | + "p0_entities": { |
| 38 | + "requirement": "Cover event, state, memory, and summary entities.", |
| 39 | + "entities": ["event", "state", "memory", "summary"], |
| 40 | + "test": "test_snapshot_contains_all_entities" |
| 41 | + }, |
| 42 | + "p0_summary_defects": { |
| 43 | + "requirement": "Detect summary loss, stale overwrite, and wrong session ownership.", |
| 44 | + "summary_loss_detection_rate": 1.0, |
| 45 | + "summary_overwrite_detection_rate": 1.0, |
| 46 | + "summary_owner_error_detection_rate": 1.0, |
| 47 | + "test": "test_summary_defect_detection" |
| 48 | + }, |
| 49 | + "p0_mutation_score": { |
| 50 | + "requirement": "Detect all registered public mutations.", |
| 51 | + "mutation_detection_rate": 1.0, |
| 52 | + "survived_mutations": [], |
| 53 | + "test": "test_mutation_detection" |
| 54 | + }, |
| 55 | + "p1_runtime_fault_detection": { |
| 56 | + "requirement": "Detect an after-commit client retry fault as a runtime replay inconsistency.", |
| 57 | + "runtime_fault_detection_rate": 1.0, |
| 58 | + "test": "test_runtime_fault_retry_duplicate_is_detected" |
| 59 | + }, |
| 60 | + "p1_persistent_rebuild": { |
| 61 | + "requirement": "Destroy and rebuild the persistent backend between replay phases, then read back through a fresh service.", |
| 62 | + "cases": [ |
| 63 | + "summary_create_update", |
| 64 | + "single_turn_text" |
| 65 | + ], |
| 66 | + "test": "test_sqlite_destroy_rebuild_continue_and_readback" |
| 67 | + }, |
| 68 | + "p0_reports": { |
| 69 | + "requirement": "Generate structured JSON diff reports with precise location fields.", |
| 70 | + "required_fields": [ |
| 71 | + "case_id", |
| 72 | + "backend_pair", |
| 73 | + "session_id", |
| 74 | + "entity_type", |
| 75 | + "entity_id", |
| 76 | + "index", |
| 77 | + "field_path", |
| 78 | + "reference_value", |
| 79 | + "actual_value", |
| 80 | + "allowed", |
| 81 | + "category", |
| 82 | + "reason" |
| 83 | + ], |
| 84 | + "test": "test_diff_report_schema_contains_required_location_fields" |
| 85 | + }, |
| 86 | + "p0_lightweight_runtime": { |
| 87 | + "requirement": "Run default lightweight mode in less than 30 seconds without external services.", |
| 88 | + "lightweight_duration_seconds_max": 30, |
| 89 | + "test": "test_lightweight_replay_cases" |
| 90 | + } |
| 91 | + } |
| 92 | +} |
0 commit comments