{ "application_manifest_sha256": "39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165", "final_quality_status": "Final V18 frozen. All 250 paired attempts and isolated E6 complete. Primary targets fail. Full E1 predicate coverage is incomplete. No V19.", "reference_root": "src/institutional_kb/", "test_root": "tests/", "findings": [ { "id": "R01", "title": "Keep the complete selected group", "status": "partially_fixed", "problem": "A keyless follow-up could accept an unrelated 9,000,000 scalar. Empty Gold membership could be replaced by a populated CT group.", "root_cause": "Fallback rerouted an isolated follow-up and could discard independent source membership. Empty-cohort rescue used the populated leg without sufficient ownership proof. An eligible CT fallback also bypassed final-answer shape validation when promoted from a merge input.", "fix": "Carry accepted route and exact membership. Retain independent Gold predicates unless the narrow CT recency rescue proves they can be superseded. Require a complete user_id cohort before an explicit snapshot money source change. Recheck actual executed SQL against the final answer shape before either eligible CT-only promotion.", "regression_tests": [ "test_answer_contract_role_promotion.py", "test_churn_tier_routing_and_merge_rescue.py", "test_federated_semantic_refusal.py", "test_insights_federated_followup_contract.py", "test_insights_v16_rank_words.py", "test_insights_v17_rank_thresholds.py", "test_prd008_federated_planner.py", "tests/test_insights_v18_membership_output.py" ], "references": [ "services/bi_federated_execution_service.py:260", "services/bi_federated_execution_service.py:1796", "services/bi_federated_execution_service.py:1832", "services/bi_federated_execution_service.py:1946", "services/bi_federated_execution_service.py:2343", "services/bi_federated_planner.py:1747" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "The proof covers supported membership routes. Held failures show that arbitrary generated source predicates are not solved. Final limit: Partially fixed. Native controls preserve 45 checked clients. Held dependent turns can miss the checked-group reference and use the wrong source or group.", "final_result": "Partially fixed. Native controls preserve 45 checked clients. Held dependent turns can miss the checked-group reference and use the wrong source or group.", "current_evidence": "V18 native intersection and union controls preserve requested aliases, ordering before capping, and complete internal member keys. Native listing, scoped total and count keep the same 45 clients.", "priority": "P1", "user_impact": "A keyless follow-up could accept an unrelated 9,000,000 scalar. Empty Gold membership could be replaced by a populated CT group.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "tests/v7/receipt.json", "review/v7-fixture-cleanup/v7-broader-classification.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Original R01 counterexamples retained. Post-V5 review added source/role/promotion controls; V7 also closes the failed-leg semantic-refusal classification residual. No failed historical evaluation arm is reclassified.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "The 9,000,000 whole-book fallback and empty authoritative Gold-group cases now refuse ineligible source loss. Eligible CT-only recovery controls pass. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "services/bi_federated_execution_service.py:256", "services/bi_federated_execution_service.py:1768", "services/bi_federated_execution_service.py:1806", "services/bi_federated_execution_service.py:2302", "services/bi_federated_planner.py:1754" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json" ], "limits": "General interpretation still depends on the live-model quality results." }, "source_reference_evidence": [ { "reference": "src/institutional_kb/services/bi_federated_execution_service.py:260", "relevance": "Reject independent Gold filters/joins before discarding the source.", "line_text": "def _gold_leg_can_be_discarded(sql: str, *, ct_owns_recency: bool = False) -> bool:", "sha256": "4b7741ed1dbf958295c3e092808e8123e3341ef082c94a128ce0d6ae472c38b1", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/services/bi_federated_execution_service.py:1796", "relevance": "First eligible keyless branch invokes final-answer promotion.", "line_text": "return self._promote_ct_result_to_answer(ct_result, request)", "sha256": "4b7741ed1dbf958295c3e092808e8123e3341ef082c94a128ce0d6ae472c38b1", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/services/bi_federated_execution_service.py:1832", "relevance": "Second eligible keyless branch invokes the same promotion gate.", "line_text": "return self._promote_ct_result_to_answer(ct_result, request)", "sha256": "4b7741ed1dbf958295c3e092808e8123e3341ef082c94a128ce0d6ae472c38b1", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/services/bi_federated_execution_service.py:1946", "relevance": "Check actual executed SQL and captured context before display cap/provenance.", "line_text": "def _promote_ct_result_to_answer(", "sha256": "4b7741ed1dbf958295c3e092808e8123e3341ef082c94a128ce0d6ae472c38b1", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/services/bi_federated_execution_service.py:2343", "relevance": "Empty-result rescue retains independent membership obligations.", "line_text": "def _rescue_annihilated_cohort(", "sha256": "4b7741ed1dbf958295c3e092808e8123e3341ef082c94a128ce0d6ae472c38b1", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/services/bi_federated_planner.py:1747", "relevance": "Routing takes accepted_cohort; explicit snapshot source change requires complete user_id membership.", "line_text": "def decide_routing(", "sha256": "cc162abb1ce72935c46968f301768999d3af4382da83490174ef5add594c75f1", "matches_frozen_v7": true } ], "test_evidence": [ { "file": "tests/test_churn_tier_routing_and_merge_rescue.py", "sha256": "6b52cf95fdea4b54eea0f90cbec201567748501012f364e701a5a214cb80819a", "v7_affected_counts": { "passed": 25 }, "exact_test_references": [ "tests/test_churn_tier_routing_and_merge_rescue.py:200 (test_empty_authoritative_acquisition_cohort_is_never_replaced)", "tests/test_churn_tier_routing_and_merge_rescue.py:231 (test_churn_rescue_preserves_other_sources_membership_obligations)" ] }, { "file": "tests/test_insights_federated_followup_contract.py", "sha256": "23cd5ed6ad0db23047dfa1cebff5721c9eaff3e9e78233a8128209d80d4e2563", "v7_affected_counts": { "passed": 18 }, "exact_test_references": [ "tests/test_insights_federated_followup_contract.py:57 (test_gold_generation_and_repair_keep_scope_and_exact_members)", "tests/test_insights_federated_followup_contract.py:98 (test_unproved_membership_stops_before_dispatch)" ] }, { "file": "tests/test_prd008_federated_planner.py", "sha256": "08221c9845a5b4a05d7301394c69ffd88f0fda8eb7ed1da47cc940c4161eac2d", "v7_affected_counts": { "passed": 140 }, "exact_test_references": [ "tests/test_prd008_federated_planner.py:1022 (test_snapshot_money_source_change_retains_exact_accepted_membership)", "tests/test_prd008_federated_planner.py:1053 (test_snapshot_money_source_change_refuses_without_complete_client_membership)", "tests/test_prd008_federated_planner.py:1075 (test_snapshot_money_source_change_cannot_drop_current_live_conditions)" ] }, { "file": "tests/test_answer_contract_role_promotion.py", "sha256": "0f2439f4d476ab1ac6e940f4d449bab7ef82fbe4fc7246ea98aad6763f446a58", "v7_affected_counts": { "passed": 6 }, "exact_test_references": [ "tests/test_answer_contract_role_promotion.py:56 (test_ct_fallback_rechecks_actual_final_shape_without_extra_execution)" ] }, { "file": "tests/test_federated_semantic_refusal.py", "sha256": "1b8ac1d1f9b99981013f4ef1913e9a79d0c15047fe3c227a6169af44adae65cc", "v7_affected_counts": { "passed": 9 }, "exact_test_references": [ "tests/test_federated_semantic_refusal.py:66 (test_ct_semantic_refusal_survives_federation_while_transport_is_unchanged)" ] } ], "historical_v7": { "status": "fixed_in_curated_checks", "final_result": "fixed_in_curated_checks", "current_evidence": "The unrelated 9,000,000 scalar and empty authoritative Gold cohort remain rejected as source-loss recovery. Complete and incomplete accepted-cohort cases, exact empty membership, current CT predicate preservation, valid CT-only recovery, and both promotion branches pass. Already classified semantic refusals retain their corrective contract through CT-only, CT-first, Gold-first and replay boundaries. V7 final affected execution: 2,234 passed in 74 files, no failures or skips. Broader: 3,704 passed, 10 failures matching the retained original baseline, 13 skipped in 184 files. Mypy: 1,166 source files clean; Ruff clean. These are structural/curated checks; no V7 model-answer outcome is asserted here.", "verification_snapshot": "final-contract-v7", "current_source_verification": "Every application reference in this suggestion was compared byte-for-byte with frozen final-contract-v7; every cited test file matches the V7 runner test manifest. V7 final affected execution: 2,234 passed in 74 files, no failures or skips. Broader: 3,704 passed, 10 failures matching the retained original baseline, 13 skipped in 184 files. Mypy: 1,166 source files clean; Ruff clean. These are structural/curated checks; no V7 model-answer outcome is asserted here." }, "historical_pre_v13": { "status": "reopened", "current_evidence": "Frozen V10 broad checks reproduce a critical inherited-fanout failure: a keyless fallback can return a whole-book9,000,000 value for a previously selected group. V11 requires both fresh CT-only meaning and carried-route agreement before promotion. The earlier red controls remain retained; final repair verification is pending.", "final_result": "unverified_on_final_candidate", "verification_snapshot": "opened_final_contract_v7_new_work_not_final" }, "v13_native_evidence": "Correct native45-client listing,2454.50total and45count;30final reads dispatch no execution. Blocked parent offers minimal recovery.", "historical_before_v17_lead_update": { "current_evidence": "V12 native three-turn listing\u2192total\u2192count preserves all45 checked client IDs and returns2454.50 then45. V13 repairs explicit ranked-book scope release; live development still has accepted-cohort planning failures.", "final_result": "unverified_on_final_candidate", "verification_snapshot": "V12 measured; V13 working scoped checks, final unverified", "limits": "Source-discard proof is deliberately narrow and exact membership must be available. Supported fallback/route controls do not prove arbitrary model-generated cohort predicates or every natural-language source request. Canned dispatch controls establish integration behavior; independent database/oracle evidence remains separate.", "root_cause": "Fallback rerouted an isolated follow-up and could discard independent source membership. Empty-cohort rescue used the populated leg without sufficient ownership proof. An eligible CT fallback also bypassed final-answer shape validation when promoted from a merge input.", "fix": "Carry accepted route and exact membership. Retain independent Gold predicates unless the narrow CT recency rescue proves they can be superseded. Require a complete user_id cohort before an explicit snapshot money source change. Recheck actual executed SQL against the final answer shape before either eligible CT-only promotion." }, "history_before_v18_boundary_diagnosis": { "current_evidence": "V16 threshold wording capped seven selected clients at one displayed row. V17 passes 165 rank and threshold controls, including generated, approved and rerun native routes." }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V16 threshold wording capped seven selected clients at one displayed row. V17 passes 165 rank and threshold controls, including generated, approved and rerun native routes. V17 live development: 20/70 correct, 7/28 first-turn correct, 1/28 first-turn error, 0/14 full conversations. Final independent evaluation remains closed. V18 native intersection reproduces V17 final alias refusal and verifies requested client_id/name, display ordering before capping, and complete checked carry. Missing name remains a refusal.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Source-discard proof is deliberately narrow and exact membership must be available. Supported fallback/route controls do not prove arbitrary model-generated cohort predicates or every natural-language source request. Canned dispatch controls establish integration behavior; independent database/oracle evidence remains separate.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "R02", "title": "Prove row identity before summing", "status": "partially_fixed", "problem": "One deal worth 100 joined to two notes could become 200. Equal values did not identify equal entities.", "root_cause": "Aggregation used repeated numeric values without establishing physical source identity or join cardinality; equal values do not identify equal entities.", "fix": "Derive bounded source-key, join-cardinality and aggregation contracts from physical lineage. Reject repeated/missing physical keys and unproved deal SUM joins. Keep separate equal-valued entities and apply only declared additive or denominator/member-aware reductions.", "regression_tests": [ "test_insights_federated_followup_contract.py", "test_insights_postgres_values.py", "test_insights_query_correctness.py", "test_insights_v17_distinct_snapshot_sum.py", "test_prd004_federated_merge.py", "tests/test_insights_v18_distinct_membership_grain.py" ], "references": [ "services/bi_federated_merge.py:844", "services/bi_federated_merge.py:883", "services/bi_federated_merge.py:903", "bi/measure_semantics.py:214", "bi/measure_semantics.py:283", "bi/measure_semantics.py:418" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "This is finite uniqueness and lineage proof. It does not validate every SQL aggregate or join. Final limit: Partially fixed. Curated grain controls pass. A scorer-correct count uses note person identity where deal person identity is required.", "final_result": "Partially fixed. Curated grain controls pass. A scorer-correct count uses note person identity where deal person identity is required.", "current_evidence": "V18 recognizes a proved complete DISTINCT key tuple. Partial tuples and set-returning projections still refuse. Native deal sums pass with distinct members and equal-valued entities.", "priority": "P1", "user_impact": "One deal worth 100 joined to two notes could become 200. Equal values did not identify equal entities.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "tests/v7/receipt.json", "review/v7-fixture-cleanup/v7-broader-classification.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Original entity-identity counterexamples remain checked on V7; the row does not promote bounded deal/merge checks into a universal SQL interpreter.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "One physical deal duplicated by notes refuses unresolved multiplication. Two distinct equal-value deals remain 200. Rates, repeated attributes, IDs and distinct aggregation controls pass. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "services/bi_federated_merge.py:883", "services/bi_federated_merge.py:903", "bi/measure_semantics.py:224", "bi/measure_semantics.py:411" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json" ], "limits": "The verifier supports a bounded set of SQL shapes. Unknown grain refuses." }, "source_reference_evidence": [ { "reference": "src/institutional_kb/services/bi_federated_merge.py:844", "relevance": "Reduce declared measures after stamping aggregation and validating source grain.", "line_text": "def reduce_gold_leg_to_key_grain(", "sha256": "aecdbf5c26b0b1079071ce02c1b5695cf451bb6c1d622d4e6f677e17de3790a9", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/services/bi_federated_merge.py:883", "relevance": "Reject missing or duplicated source identity.", "line_text": "def _validate_source_grain(result: \"QueryExecutionResult\") -> None:", "sha256": "aecdbf5c26b0b1079071ce02c1b5695cf451bb6c1d622d4e6f677e17de3790a9", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/services/bi_federated_merge.py:903", "relevance": "Apply declared aggregate operation and denominator/member requirements.", "line_text": "def _aggregate_value(result: \"QueryExecutionResult\", idx: int, rows: list[list[Any]]) -> Any:", "sha256": "aecdbf5c26b0b1079071ce02c1b5695cf451bb6c1d622d4e6f677e17de3790a9", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/bi/measure_semantics.py:214", "relevance": "Resolve positional scope projections before physical lineage.", "line_text": "def _scope_projection(name: str, scope: Scope) -> exp.Expression | None:", "sha256": "244a5297fed632cd993c88d36a01bbc9c5ab18dc207e358f2959e888336836f7", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/bi/measure_semantics.py:283", "relevance": "Check whether joins preserve the measure source.", "line_text": "def _joins_preserve_source(column: exp.Column, scope: Scope) -> bool:", "sha256": "244a5297fed632cd993c88d36a01bbc9c5ab18dc207e358f2959e888336836f7", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/bi/measure_semantics.py:418", "relevance": "Bounded pre-execution reachable deal-SUM grain rule.", "line_text": "def additive_grain_violations(sql: str) -> tuple[str, ...]:", "sha256": "244a5297fed632cd993c88d36a01bbc9c5ab18dc207e358f2959e888336836f7", "matches_frozen_v7": true } ], "test_evidence": [ { "file": "tests/test_insights_query_correctness.py", "sha256": "438c2614733c88189c3faf43de913d9f7293ca7946292b927f824d0d8eced802", "v7_affected_counts": { "passed": 96 }, "exact_test_references": [ "tests/test_insights_query_correctness.py:287 (test_single_source_reachable_deal_sum_requires_join_grain_proof)", "tests/test_insights_query_correctness.py:366 (test_joined_deal_identity_distinguishes_duplicates_from_equal_distinct_deals)", "tests/test_insights_query_correctness.py:436 (test_deal_identity_cannot_be_borrowed_from_another_cte_instance)", "tests/test_insights_query_correctness.py:503 (test_rate_uses_defined_denominator_and_unknown_rates_refuse)" ] }, { "file": "tests/test_insights_postgres_values.py", "sha256": "acc1a1682f892cc104a44e5831a927f7163bbba271485d707f5534c54ac250e5", "v7_affected_counts": { "passed": 22 }, "exact_test_references": [ "tests/test_insights_postgres_values.py:155 (test_executed_deal_join_matches_independent_entity_total)" ] }, { "file": "tests/test_insights_federated_followup_contract.py", "sha256": "23cd5ed6ad0db23047dfa1cebff5721c9eaff3e9e78233a8128209d80d4e2563", "v7_affected_counts": { "passed": 18 }, "exact_test_references": [ "tests/test_insights_federated_followup_contract.py:179 (test_final_sql_boundary_rejects_multiplied_deal_sum)" ] }, { "file": "tests/test_prd004_federated_merge.py", "sha256": "c4cb40e497cf5778d7afb2446277e1d929f926fc2f6a929cd438684be76b8f15", "v7_affected_counts": { "passed": 26 }, "exact_test_references": [] } ], "historical_pre_v13": { "status": "fixed_in_curated_checks", "current_evidence": "A duplicated physical deal refuses unresolved multiplication, while two distinct 100-valued deals total 200. Final executable SQL rejects an unproved deal SUM join; positional aliases, separate CTE instances, verified unique/preaggregated joins, repeated attributes, defined ratios and distinct/nonadditive controls pass. V7 final affected execution: 2,234 passed in 74 files, no failures or skips. Broader: 3,704 passed, 10 failures matching the retained original baseline, 13 skipped in 184 files. Mypy: 1,166 source files clean; Ruff clean. These are structural/curated checks; no V7 model-answer outcome is asserted here.", "final_result": "fixed_in_curated_checks", "verification_snapshot": "final-contract-v7" }, "v13_native_evidence": "Separate duplicate-trade fixture produces explicit grain refusal with real guard, code, reason and rephrase. No error projection is installed.", "historical_before_v17_lead_update": { "current_evidence": "The V12 duplicate-grain stimulus was disproved: its56 rows had56 distinct keys. The separate V14 native fixture has a repeated key and triggers the actual grain guard. The outward refusal is correct, but nested outcome parity failed; V15 repairs that representation.", "final_result": "Final V15 verification and independent acceptance pending.", "verification_snapshot": "V12 measured; V13 working scoped checks, final unverified", "limits": "The pre-execution grain rule covers reachable deal SUM scopes and supported lineage, not all possible SQL measures/joins. An empty violation list is not universal grain proof. Merge re-aggregation refuses unknown semantics when combination is required; one-row passthrough is not a general aggregation certificate.", "root_cause": "Aggregation used repeated numeric values without establishing physical source identity or join cardinality; equal values do not identify equal entities.", "fix": "Derive bounded source-key, join-cardinality and aggregation contracts from physical lineage. Reject repeated/missing physical keys and unproved deal SUM joins. Keep separate equal-valued entities and apply only declared additive or denominator/member-aware reductions." }, "history_before_v18_boundary_diagnosis": { "current_evidence": "V17 one-per-client DISTINCT identity proof preserves two equal-valued clients. Amount DISTINCT and foreign identity remain rejected. The broader gate passes 270 controls." }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V17 one-per-client DISTINCT identity proof preserves two equal-valued clients. Amount DISTINCT and foreign identity remain rejected. The broader gate passes 270 controls. V17 live development: 20/70 correct, 7/28 first-turn correct, 1/28 first-turn error, 0/14 full conversations. Final independent evaluation remains closed. V17 rejected a DISTINCT client-key CTE joined to deal owners. V18 adds complete-tuple uniqueness proof; partial tuple and set-returning controls remain required.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "The pre-execution grain rule covers reachable deal SUM scopes and supported lineage, not all possible SQL measures/joins. An empty violation list is not universal grain proof. Merge re-aggregation refuses unknown semantics when combination is required; one-row passthrough is not a general aggregation certificate.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "R03", "title": "Follow physical column lineage", "status": "partially_fixed", "problem": "Positional outputs could rename category to status. PostgreSQL quoted Status or Clients could be confused with the required lowercase physical identifier.", "root_cause": "Predicate matching lost positional-output lineage and conflated PostgreSQL case-sensitive quoted physical columns/tables with lowercase identifiers.", "fix": "Normalize only unquoted PostgreSQL identifiers. Resolve CTE/derived aliases and positional outputs to physical columns before comparing required root/schema/table/column identity. Return unverified for unsupported quoted physical shapes.", "regression_tests": [ "test_insights_quoted_semantic_identity.py", "test_insights_postgres_values.py", "test_insights_query_correctness.py", "test_insights_v17_distinct_snapshot_sum.py" ], "references": [ "bi/semantic_obligations.py:51", "bi/semantic_obligations.py:70", "bi/semantic_obligations.py:94", "bi/measure_semantics.py:111", "bi/sql_source_validation.py:44" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Semantic-obligation metadata is not universal execution authorization or whole-question proof. Final limit: Partially fixed. Curated physical-lineage controls pass. Held query generation still drops filters and chooses incorrect entity keys.", "final_result": "Partially fixed. Curated physical-lineage controls pass. Held query generation still drops filters and chooses incorrect entity keys.", "current_evidence": "Quoted identity, physical lineage, DISTINCT snapshot sums, and V18 complete-tuple controls pass on the final source.", "priority": "P1", "user_impact": "Positional outputs could rename category to status. PostgreSQL quoted Status or Clients could be confused with the required lowercase physical identifier.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "tests/v7/receipt.json", "review/v7-fixture-cleanup/v7-broader-classification.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "The physical-identity repair is retained from the prior independent PostgreSQL counterexamples and rechecked in the V7 affected run. Replace historical aggregate counts with explicit current module counts, rather than implying 36/17 are this module alone.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "Positional category-to-status proof is unverified. Independent PG member sets distinguish lowercase status member2 from quoted Status member1. Harmless CTE/alias controls remain satisfied. 19 pre-fix quoted-identity failures are retained. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "bi/semantic_obligations.py:51", "bi/semantic_obligations.py:70", "bi/semantic_obligations.py:94", "bi/measure_semantics.py:110" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json", "review/final-query-review-v5.md" ], "limits": "36 focused checks include 17 actual PostgreSQL cases. Safe lowercase quotes, unquoted case folding and quoted derived aliases remain supported. The verifier cannot prove arbitrary natural-language meaning." }, "source_reference_evidence": [ { "reference": "src/institutional_kb/bi/semantic_obligations.py:51", "relevance": "Canonical predicate comparison resolves physical lineage before name comparison.", "line_text": "def _canonical(", "sha256": "e79607be23b5c4e1762dfad780c88c4239f012b3fc2d309ca31b6c2f682bff50", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/bi/semantic_obligations.py:70", "relevance": "Simple-path proof preserves physical schema/table identity and rejects unsupported shapes.", "line_text": "def _simple_path(scope: Scope) -> list[tuple[Scope, str]] | None:", "sha256": "e79607be23b5c4e1762dfad780c88c4239f012b3fc2d309ca31b6c2f682bff50", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/bi/semantic_obligations.py:94", "relevance": "Return satisfied/violated/unverified obligations; question interpretation remains unverified.", "line_text": "def verify_semantic_obligations(", "sha256": "e79607be23b5c4e1762dfad780c88c4239f012b3fc2d309ca31b6c2f682bff50", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/bi/measure_semantics.py:111", "relevance": "Resolve columns through physical source and derived output lineage.", "line_text": "def _resolve_column_base(", "sha256": "244a5297fed632cd993c88d36a01bbc9c5ab18dc207e358f2959e888336836f7", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/bi/sql_source_validation.py:44", "relevance": "Apply PostgreSQL folding only to unquoted identifiers.", "line_text": "def normalize_postgres_identifiers(tree: exp.Expr) -> None:", "sha256": "93b903213a61c4977703dbb6b1c4532d5156e586b91bae46c35a9745fd308185", "matches_frozen_v7": true } ], "test_evidence": [ { "file": "tests/test_insights_quoted_semantic_identity.py", "sha256": "5a688e685ec3208c74c39085b5c48c0b75c8626c00c565d4c6351d65473839d2", "v7_affected_counts": { "passed": 31 }, "exact_test_references": [ "tests/test_insights_quoted_semantic_identity.py:68 (test_authorized_quoted_physical_names_do_not_gain_semantic_proof)", "tests/test_insights_quoted_semantic_identity.py:83 (test_physical_definition_schema_table_and_column_names_are_not_conflated)", "tests/test_insights_quoted_semantic_identity.py:112 (test_quoted_physical_proof_matches_independent_postgres_membership)" ] }, { "file": "tests/test_insights_postgres_values.py", "sha256": "acc1a1682f892cc104a44e5831a927f7163bbba271485d707f5534c54ac250e5", "v7_affected_counts": { "passed": 22 }, "exact_test_references": [ "tests/test_insights_postgres_values.py:118 (test_positional_alias_proof_agrees_with_physical_membership)" ] }, { "file": "tests/test_insights_query_correctness.py", "sha256": "438c2614733c88189c3faf43de913d9f7293ca7946292b927f824d0d8eced802", "v7_affected_counts": { "passed": 96 }, "exact_test_references": [ "tests/test_insights_query_correctness.py:553 (test_semantic_obligations_distinguish_proof_omission_and_uncertainty)" ] } ], "status_before_current_final_verification": "fixed_in_curated_checks", "historical_before_v17_lead_update": { "current_evidence": "V7 has 31 passing cases in test_insights_quoted_semantic_identity.py, including 13 actual PostgreSQL member-set cases. Required lowercase status selects member 2; quoted Status, quoted Clients and a dotted quoted table select different independently asserted members and cannot receive satisfied proof. Lowercase quotes, unquoted uppercase folding and harmless quoted/positional derived aliases remain supported. The separate positional category/status PostgreSQL test also passes. V7 final affected execution: 2,234 passed in 74 files, no failures or skips. Broader: 3,704 passed, 10 failures matching the retained original baseline, 13 skipped in 184 files. Mypy: 1,166 source files clean; Ruff clean. These are structural/curated checks; no V7 model-answer outcome is asserted here.", "final_result": "fixed_in_curated_checks", "verification_snapshot": "final-contract-v7", "limits": "An unverified label does not itself refuse execution: this helper emits semantic-obligation metadata. It does not prove the whole question, authorize sources, or itself block every unsupported query. The quoted-identity SQL tests deliberately authorize both distinct names so semantic proof is tested independently of authorization. Only supported simple single-input predicate paths can be certified.", "root_cause": "Predicate matching lost positional-output lineage and conflated PostgreSQL case-sensitive quoted physical columns/tables with lowercase identifiers.", "fix": "Normalize only unquoted PostgreSQL identifiers. Resolve CTE/derived aliases and positional outputs to physical columns before comparing required root/schema/table/column identity. Return unverified for unsupported quoted physical shapes." }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "The retained V16 DISTINCT ON client-key SUM was refused despite valid physical snapshot lineage. V17 scoped proof and native SQL return the independently checked GBP 1127.80.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "An unverified label does not itself refuse execution: this helper emits semantic-obligation metadata. It does not prove the whole question, authorize sources, or itself block every unsupported query. The quoted-identity SQL tests deliberately authorize both distinct names so semantic proof is tested independently of authorization. Only supported simple single-input predicate paths can be certified.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "R04", "title": "Use the captured parent", "status": "partially_fixed", "problem": "Helpers could load a later turn after a worker captured an older parent. An explicit root could acquire old scope.", "root_cause": "Context helpers could reload later thread turns instead of captured ancestry; explicit-null roots and omitted parents shared a fallback. Exact saved SQL reruns could also inherit a newer parent or changed request question.", "fix": "Resolve only valid captured-parent ancestry. Preserve explicit roots. For exact saved SQL or merge replay, pin the saved question/recognizer/resolution and saved source choices, and derive no new follow-up/cohort context from a later parent. Keep ordinary follow-ups and confirmations on their captured parent.", "regression_tests": [ "test_sql_anchor_turn_polluted_thread.py", "test_insights_v8_chain_recovery.py" ], "references": [ "repositories/bi_conversation.py:653", "api/routers/bi_query.py:7791", "api/routers/bi_query.py:7822", "api/routers/bi_query.py:7844", "api/routers/bi_query.py:8040" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Arbitrary branch language and full live conversation correctness remain separate. Final limit: Partially fixed. Captured-parent tests pass, but all 44 explicit continuations after failed/refused parents miss the minimal recovery gate.", "final_result": "Partially fixed. Captured-parent tests pass, but all 44 explicit continuations after failed/refused parents miss the minimal recovery gate.", "current_evidence": "The 57 final chain controls use the captured parent. Fresh native listing, scoped total and count retain checked membership.", "priority": "P1", "user_impact": "Helpers could load a later turn after a worker captured an older parent. An explicit root could acquire old scope.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "tests/v7/receipt.json", "review/v7-fixture-cleanup/v7-broader-classification.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Captured-ancestry repairs retained; post-V5 exact-rerun context and saved-choice fixes now have V7 integrated verification. Historical V5 outcome remains failed.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "partially_fixed", "current_evidence": "Older successful/clarification parents, explicit roots, branches, invalid ancestry, scope release and concurrent completion retain their captured ancestry. Delayed full-thread GET regression is covered in the browser suite. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "repositories/bi_conversation.py:653", "api/routers/bi_query.py:7761", "api/routers/bi_query.py:9605" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json" ], "limits": "Branch and concurrent cases are fixture based. Live multi-turn success is measured separately." }, "source_reference_evidence": [ { "reference": "src/institutional_kb/repositories/bi_conversation.py:653", "relevance": "Explicit-root handling and ancestry validation exclude siblings/later completions.", "line_text": "def list_ancestor_turns(", "sha256": "2bc8939f9d051b31cd9c8bbb83027cf705d254dafa6ea72b071770272cc5b70b", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/api/routers/bi_query.py:7791", "relevance": "Resolve saved SQL/replay only from the current conversation.", "line_text": "if body.rerun_turn_id is not None:", "sha256": "afc564ebf7aa73c48e730e72479eaca4bf22d2f813e4be7edffd5fbe655da2c2", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/api/routers/bi_query.py:7822", "relevance": "Pin exact rerun question and recognition/resolution to the saved source.", "line_text": "if exact_rerun and rerun_source is not None:", "sha256": "afc564ebf7aa73c48e730e72479eaca4bf22d2f813e4be7edffd5fbe655da2c2", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/api/routers/bi_query.py:7844", "relevance": "Exact rerun has no new context parent.", "line_text": "context_parent = None if exact_rerun else previous_turn", "sha256": "afc564ebf7aa73c48e730e72479eaca4bf22d2f813e4be7edffd5fbe655da2c2", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/api/routers/bi_query.py:8040", "relevance": "Restore saved source bindings and prevent inherited bindings from replacing them.", "line_text": "if rerun_source is not None and (rerun_sql is not None or rerun_merge_replay is not None):", "sha256": "afc564ebf7aa73c48e730e72479eaca4bf22d2f813e4be7edffd5fbe655da2c2", "matches_frozen_v7": true } ], "test_evidence": [ { "file": "tests/test_sql_anchor_turn_polluted_thread.py", "sha256": "64be3930999f2abf6fb00f963b375dc9f684d6590ee889f399182db0a647f0e8", "v7_affected_counts": { "passed": 8 }, "exact_test_references": [ "tests/test_sql_anchor_turn_polluted_thread.py:87 (test_explicit_root_does_not_inherit_a_populated_thread)", "tests/test_sql_anchor_turn_polluted_thread.py:99 (test_older_clarification_cannot_anchor_on_a_later_answer)", "tests/test_sql_anchor_turn_polluted_thread.py:124 (test_captured_branch_ancestry_excludes_other_roots_and_later_turns)" ] }, { "file": "tests/test_bic015_async_turn_jobs.py", "sha256": "0d110d3ed1202bee12c1244e5e32839fd6aa2230a55532e2ea009e9de6d54ecc", "v7_affected_counts": { "passed": 39 }, "exact_test_references": [] }, { "file": "tests/test_answer_contract_saved_rerun_context.py", "sha256": "ccd42c5cfefe35680b8d29323e348d8400fba503a4e309b011abd4ad17f7fcf9", "v7_affected_counts": { "passed": 6 }, "exact_test_references": [ "tests/test_answer_contract_saved_rerun_context.py:29 (test_saved_rerun_has_no_new_parent_cohort_or_shape_context)" ] }, { "file": "tests/test_insights_answer_contract_execution.py", "sha256": "9d3318b456c3cc3fd5bda3d0002e89b9a477357be7708c4ddc07a2a1b553b03e", "v7_affected_counts": { "passed": 9 }, "exact_test_references": [ "tests/test_insights_answer_contract_execution.py:133 (test_replay_dispatch_preserves_saved_measure_choice)" ] } ], "historical_v7": { "status": "fixed_in_curated_checks", "final_result": "fixed_in_curated_checks", "current_evidence": "V7 passes eight SQL-anchor ancestry tests, 39 async-job tests and six saved-rerun context cases. Saved clients 1/2 cannot acquire newer parent client 99; unavailable newer membership cannot force an exact rerun clarification. A changed submitted question cannot relabel saved SQL meaning. Saved source choices survive approved/replay dispatch. V7 final affected execution: 2,234 passed in 74 files, no failures or skips. Broader: 3,704 passed, 10 failures matching the retained original baseline, 13 skipped in 184 files. Mypy: 1,166 source files clean; Ruff clean. These are structural/curated checks; no V7 model-answer outcome is asserted here.", "verification_snapshot": "final-contract-v7", "current_source_verification": "Every application reference in this suggestion was compared byte-for-byte with frozen final-contract-v7; every cited test file matches the V7 runner test manifest. V7 final affected execution: 2,234 passed in 74 files, no failures or skips. Broader: 3,704 passed, 10 failures matching the retained original baseline, 13 skipped in 184 files. Mypy: 1,166 source files clean; Ruff clean. These are structural/curated checks; no V7 model-answer outcome is asserted here." }, "historical_pre_v14_result_review": { "status": "reopened", "root_cause": "Context helpers could reload later thread turns instead of captured ancestry; explicit-null roots and omitted parents shared a fallback. Exact saved SQL reruns could also inherit a newer parent or changed request question.", "fix": "Resolve only valid captured-parent ancestry. Preserve explicit roots. For exact saved SQL or merge replay, pin the saved question/recognizer/resolution and saved source choices, and derive no new follow-up/cohort context from a later parent. Keep ordinary follow-ups and confirmations on their captured parent.", "current_evidence": "Opened V7 scope release misses \u201cRelease the previous client selection\u201d plus \u201call distinct CT snapshot clients\u201d. The top-booked b4 response repeats prior currency pipeline amounts instead of the requested all-client count. New root/missing-parent and release regressions reproduced10 failures before the next fix.", "regression_tests": [ "test_sql_anchor_turn_polluted_thread.py", "test_bic015_async_turn_jobs.py", "test_answer_contract_saved_rerun_context.py", "test_insights_answer_contract_execution.py" ], "limits": "These are deterministic repository/endpoint/job controls, with canned dispatch where stated. They do not measure arbitrary live multi-turn interpretation or certify unreviewed browser timing. Legacy turns without recorded parent edges use documented predecessor inference.", "final_result": "reopened" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_sql_anchor_turn_polluted_thread.py", "test_insights_v8_chain_recovery.py" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "Captured-parent controls forbid older, later or unrelated group reuse. V15 native 45-client listing -> total -> count retains the checked keys. The separate pure-count endpoint closes the observed seven-key request returning three rows.", "final_result": "V15_frozen_curated_evidence; independent_quality_targets_unreported", "verification_snapshot": "opened_final_contract_v7_new_work_not_final", "limits": "Arbitrary branch language is not certified. Native covers a three-turn chain; deterministic controls cover three and five turns.", "root_cause": "Context helpers could reload later thread turns instead of captured ancestry; explicit-null roots and omitted parents shared a fallback. Exact saved SQL reruns could also inherit a newer parent or changed request question.", "fix": "Resolve only valid captured-parent ancestry. Preserve explicit roots. For exact saved SQL or merge replay, pin the saved question/recognizer/resolution and saved source choices, and derive no new follow-up/cohort context from a later parent. Keep ordinary follow-ups and confirmations on their captured parent." }, "v16_independent_result_disposition": { "id": "R04", "title": "Use the captured parent", "status": "fixed_on_curated_controls", "root_cause": "Context helpers could reload later thread turns instead of captured ancestry; explicit-null roots and omitted parents shared a fallback. Exact saved SQL reruns could also inherit a newer parent or changed request question.", "fix": "Resolve only valid captured-parent ancestry. Preserve explicit roots. For exact saved SQL or merge replay, pin the saved question/recognizer/resolution and saved source choices, and derive no new follow-up/cohort context from a later parent. Keep ordinary follow-ups and confirmations on their captured parent.", "current_evidence": "V16 native 45-client listing -> scoped total -> scalar count retains checked membership. Captured ancestry and the separate seven-key count endpoint have deterministic controls. The currency NULL-policy group routing bug also has a direct retained-parent reproduction and repair.", "regression_tests": [ "test_sql_anchor_turn_polluted_thread.py", "test_insights_v8_chain_recovery.py" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json", "results-v8/v16-carry-pair-diagnosis/receipt.json", "results-v8/v16-carry-pair-diagnosis/repaired-review.json" ], "remaining_limits": "Arbitrary branch language is not certified. Native covers a three-turn chain; deterministic controls cover three and five turns.", "final_result": "V16 frozen curated evidence; independent quality targets remain evaluator-owned" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V16 native 45-client listing -> scoped total -> scalar count retains checked membership. Captured ancestry and the separate seven-key count endpoint have deterministic controls. The currency NULL-policy group routing bug also has a direct retained-parent reproduction and repair.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Arbitrary branch language is not certified. Native covers a three-turn chain; deterministic controls cover three and five turns.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "R05", "title": "Keep units and non-measure types", "status": "partially_fixed", "problem": "USD 100 and EUR 200 could become an unlabeled 300 total. Numeric IDs could become headline measures.", "root_cause": "Unit lineage was lost in aggregation and rendering. Currency proof compared names without complete source/physical lineage. Some views inferred measure types from aliases before honoring metadata.", "fix": "Follow each amount through lexical CTE and derived-table lineage. Require the physical currency or rate from the same source instance. Preserve native unit bindings and authoritative types in every view, including sorted rows. Reject unresolved reduction; retain valid converted CTEs.", "regression_tests": [ "test_insights_count_metadata.py", "js/bi-answer-values.test.mjs", "test_insights_result_source_provenance.py" ], "references": [ "static/js/bi.js:2783", "static/js/bi.js:10069", "static/js/bi.js:10102", "static/js/bi.js:13464", "bi/answer_record.py:10", "bi/measure_semantics.py:570", "services/bi_query_execution_service.py:1531", "services/bi_query_execution_service.py:3775", "services/bi_insight_summary.py:170" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Mixed-currency native browser behavior remains unverified. The latency alias mismatch is retained. Final limit: Partially fixed. Native GBP scope and values agree. Mixed-currency browser output and all generated measure choices remain unverified.", "final_result": "Partially fixed. Native GBP scope and values agree. Mixed-currency browser output and all generated measure choices remain unverified.", "current_evidence": "Final checked-fact narrative and authoritative value metadata controls pass. Native GBP values and NULLs agree across the observed surfaces.", "priority": "P1", "user_impact": "USD 100 and EUR 200 could become an unlabeled 300 total. Numeric IDs could become headline measures.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "browser/ui/terminal-ready-recovery/receipt.json", "browser/ui/opened-r05/receipt.md", "browser/structural-final-v7/structural-verification.json", "tests/v7/affected.xml", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Remaining recheck gap repaired and verified on V5.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "partially_fixed", "current_evidence": "V5 curated cases pass. Opened live evaluation finds native pipeline_value missing unit lineage, and proven identifier aggregation overwritten by numeric metadata. Post-V5 repair is in progress. Post-V5: native pipeline_value unit and identifier repairs pass 229 scoped Python, 102 separate temporary-PostgreSQL, 23 Node value and four PG-to-JS helper controls. Final integrated evaluation remains pending.", "verification_snapshot": "final-lineage-v5", "references": [ "bi/measure_semantics.py:449", "bi/currency_normalization_guard.py:187", "bi/currency_normalization_guard.py:233", "bi/currency_normalization_guard.py:605", "static/js/bi.js:10070", "static/js/bi.js:13453", "services/bi_insight_summary.py:170" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json", "review/final-query-review-v5.md", "review/ui-final-challenge.md" ], "limits": "Bounded lineage proof does not certify arbitrary arithmetic or final output dimensions. No implicit FX conversion. A live case without an expected unit cannot certify currency correctness. An already-collapsed unknown-currency scalar cannot recover its units from display metadata. Pre-execution validation is required." }, "historical_pre_v14_result_review": { "status": "fixed_in_curated_checks", "root_cause": "Unit lineage was lost in aggregation and rendering. Currency proof compared names without complete source/physical lineage. Some views inferred measure types from aliases before honoring metadata.", "fix": "Follow each amount through lexical CTE and derived-table lineage. Require the physical currency or rate from the same source instance. Preserve native unit bindings and authoritative types in every view, including sorted rows. Reject unresolved reduction; retain valid converted CTEs.", "current_evidence": "V7 uses the exact tested JavaScript: mixed row currencies remain separate; authoritative IDs/non-measures are excluded from metrics and series; explicit exact-case native bindings survive sorting, reload, CSV and share; unknown/missing/duplicate bindings stay unknown; unsafe integer strings remain exact. The actual V7 invoice path preserves 50 GBP 24 rows and GBP 1,200 total across table/chart/CSV/reload/share. The post-V5 native pipeline and identifier repair has separate actual-PG metadata evidence; this audit does not elevate those older backend test counts to a new V7 suite result.", "regression_tests": [ "test_insights_query_correctness.py", "test_insights_answer_record.py", "test_bi_share_snapshot.py", "js/bi-answer-values.test.mjs", "e2e/insights-review-fixes.spec.js", "test_insights_quoted_measure_identity.py", "test_insights_quoted_currency_pins.py" ], "limits": "No actual V7 mixed-currency browser\u2192source path was added: adversarial browser APIs are mocked. The homogeneous invoice fixture cannot certify arbitrary unit lineage. Already-collapsed unknown-unit scalars require pre-execution enforcement; UI metadata cannot reconstruct their units. Backend final-source checks and live quality gates remain separately owned.", "final_result": "fixed_in_curated_checks" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_insights_count_metadata.py", "js/bi-answer-values.test.mjs", "test_insights_result_source_provenance.py" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "V15 native listing, total and count preserve 45 members and GBP 2454.50 across table, headline, CSV, authorized full-details share and reload. Native PostgreSQL -> API -> saved reload controls prove direct COUNT metadata. Current Node value/source tests pass.", "final_result": "V15_frozen_curated_evidence; independent_quality_targets_unreported", "verification_snapshot": "final-contract-v7", "limits": "Native browser money is homogeneous GBP. Mixed-unit and source adversaries have deterministic coverage. No general model precision claim.", "root_cause": "Unit lineage was lost in aggregation and rendering. Currency proof compared names without complete source/physical lineage. Some views inferred measure types from aliases before honoring metadata.", "fix": "Follow each amount through lexical CTE and derived-table lineage. Require the physical currency or rate from the same source instance. Preserve native unit bindings and authoritative types in every view, including sorted rows. Reject unresolved reduction; retain valid converted CTEs." }, "v16_independent_result_disposition": { "id": "R05", "title": "Keep units and non-measure types", "status": "partially_fixed", "root_cause": "Unit lineage was lost in aggregation and rendering. Currency proof compared names without complete source/physical lineage. Some views inferred measure types from aliases before honoring metadata.", "fix": "Follow each amount through lexical CTE and derived-table lineage. Require the physical currency or rate from the same source instance. Preserve native unit bindings and authoritative types in every view, including sorted rows. Reject unresolved reduction; retain valid converted CTEs.", "current_evidence": "V16 native listing, total and count preserve exact keys, NULLs and GBP 2454.50 across saved API, table, headline, snapshot CSV, authorized full-details share and reload. Fresh 244 Node checks preserve authoritative value/source metadata.", "regression_tests": [ "test_insights_count_metadata.py", "js/bi-answer-values.test.mjs", "test_insights_result_source_provenance.py" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json" ], "remaining_limits": "Native browser money is homogeneous GBP. Mixed-unit and source adversaries have deterministic coverage. No general model precision claim.", "final_result": "V16 frozen curated evidence; independent quality targets remain evaluator-owned" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V16 native listing, total and count preserve exact keys, NULLs and GBP 2454.50 across saved API, table, headline, snapshot CSV, authorized full-details share and reload. Fresh 244 Node checks preserve authoritative value/source metadata.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Native browser money is homogeneous GBP. Mixed-unit and source adversaries have deterministic coverage. No general model precision claim.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "R06", "title": "Recover result reads without another query", "status": "partially_fixed", "problem": "A done job followed by HTTP 503 could leave a spinner. Optional updates could replace reading state and focus.", "root_cause": "The UI treated a done job as a retrieved answer. Optional full-thread refresh could replace accepted local reading and ancestry state.", "fix": "Retry the stored result read. Show a recoverable result-ready state. Preserve the selected view, disclosures, focus and current thread state.", "regression_tests": [ "test_bic015_async_turn_jobs.py", "e2e/insights-review-fixes.spec.js" ], "references": [ "static/js/bi.js:7428", "static/js/bi.js:7466", "static/js/bi.js:7493", "static/js/bi.js:7566", "static/js/bi.js:7647", "static/js/bi.js:7910", "static/js/bi.js:8224" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "The fresh native six-case path does not inject a read outage. Final limit: Partially fixed. Final native reads start no new execution. Broader surface coverage remains incomplete.", "final_result": "Partially fixed. Final native reads start no new execution. Broader surface coverage remains incomplete.", "current_evidence": "Thirty native saved-answer reads start zero new execution. Retained source-bound Chromium controls cover an unavailable saved-result read.", "priority": "P2", "user_impact": "A done job followed by HTTP 503 could leave a spinner. Optional updates could replace reading state and focus.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "browser/ui/terminal-ready-recovery/receipt.json", "browser/ui/terminal-ready-recovery/integration-check/receipt.json", "browser/ui/actual-delayed-names/final-v7/final-receipt.json", "browser/structural-final-v7/repeated-read-source-proof.json", "tests/v7/affected.xml", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Remaining recheck gap repaired and verified on V5.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "partially_fixed", "current_evidence": "Done-plus-503 retrieval, reload, navigation and delayed update fixtures recover without rerun. Five actual selected-turn reads return identical50 rows; jobs12 and model calls11 stay unchanged. Independent PG counters remain829/0/339950. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "static/js/bi.js:7541", "static/js/bi.js:7592", "static/js/bi.js:8421", "api/routers/bi_query.py:9315" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json", "review/ui-final-challenge.md" ], "limits": "Error/race browser cases use declared API fixtures. The structural browser path uses the actual API and storage." }, "v13_native_evidence": "Correct native45-client listing,2454.50total and45count;30final reads dispatch no execution. Blocked parent offers minimal recovery.", "historical_pre_v14_result_review": { "status": "fixed_in_curated_checks", "root_cause": "The UI treated a done job as a retrieved answer. Optional full-thread refresh could replace accepted local reading and ancestry state.", "fix": "Retry the stored result read. Show a recoverable result-ready state. Preserve the selected view, disclosures, focus and current thread state.", "current_evidence": "Six cancellation/timeout regression cases failed before repair and pass afterward. The full 55-case Chromium run plus one separately added dispatched case cover done/ready GET503, repeated-read/manual retry, stopped optional stages with durable ready_at, unpublished-ready negative, stale full snapshots, navigation and focused disclosure preservation. The backend terminal state stays unchanged and retrieval adds no execution POST. V7 actual delayed names additionally preserve focused draft, receipt, table, scroll 100 and reload; five saved reads after completion leave jobs/model/cursor/PG counters unchanged.", "regression_tests": [ "e2e/insights-review-fixes.spec.js", "test_bic015_async_turn_jobs.py" ], "limits": "Error/race schedules use mocked APIs in real Chromium; actual V7 happy-path reads use real API/worker/PG/storage. No actual transport fault plus backend cancellation was injected in the V7 database harness. Cancelled and timed_out branches have direct browser controls; failed shares the branch but has no new named browser case. No universal navigation or multi-process guarantee.", "final_result": "fixed_in_curated_checks" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_bic015_async_turn_jobs.py", "e2e/insights-review-fixes.spec.js" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "V15 six native cases each permit five final GETs with no additional SQL, generation, jobs or usage. Current 56-case mocked Chromium suite verifies saved-result retry and state preservation.", "final_result": "V15_frozen_curated_evidence; independent_quality_targets_unreported", "verification_snapshot": "final-contract-v7", "limits": "Native cases do not inject a result-read outage. Fault schedules use mocked APIs in real Chromium.", "root_cause": "The UI treated a done job as a retrieved answer. Optional full-thread refresh could replace accepted local reading and ancestry state.", "fix": "Retry the stored result read. Show a recoverable result-ready state. Preserve the selected view, disclosures, focus and current thread state." }, "v16_independent_result_disposition": { "id": "R06", "title": "Recover result reads without another query", "status": "fixed_on_curated_controls", "root_cause": "The UI treated a done job as a retrieved answer. Optional full-thread refresh could replace accepted local reading and ancestry state.", "fix": "Retry the stored result read. Show a recoverable result-ready state. Preserve the selected view, disclosures, focus and current thread state.", "current_evidence": "All six V16 native cases each permit five final GETs with identical SQL, generation, job and usage counters. Fresh 56 main mocked Chromium cases cover failed saved-result read recovery and view state.", "regression_tests": [ "test_bic015_async_turn_jobs.py", "e2e/insights-review-fixes.spec.js" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json" ], "remaining_limits": "Native cases do not inject a result-read outage. Fault schedules use mocked APIs in real Chromium.", "final_result": "V16 frozen curated evidence; independent quality targets remain evaluator-owned" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "All six V16 native cases each permit five final GETs with identical SQL, generation, job and usage counters. Fresh 56 main mocked Chromium cases cover failed saved-result read recovery and view state.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Native cases do not inject a result-read outage. Fault schedules use mocked APIs in real Chromium.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "R07", "title": "Bound optional work and actual local resources", "status": "partially_fixed", "problem": "Optional labels delayed publication. Slow response bytes could outlive the job budget. A PG cancel acknowledgment could block the client.", "root_cause": "Required-ready publication followed optional naming. Inactivity/Future timeouts did not stop underlying provider resources. A cancel acknowledgment could block.", "fix": "Publish checked rows before optional presentation. Kill and reap local provider transports. Close canceled PG connections and retain server deadlines.", "regression_tests": [ "test_insights_shared_presentation_budget_review.py", "test_insights_v8_provider_attempt_deadline.py", "test_insights_pg_deadline.py" ], "references": [ "api/routers/bi_query.py:9741", "api/routers/bi_query.py:9746", "api/routers/bi_query.py:9752", "static/js/bi.js:7428", "static/js/bi.js:7589", "services/llm_executor.py:90", "services/llm_executor.py:149", "services/bi_pg_job_transport.py:19", "services/bi_ssh_job_transport.py:23" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Two DOM readiness observations are not p95. Remote resource use and cross-process capacity remain unverified. Final limit: Partially fixed. Optional 20-second work does not block the two native row observations. All 180 E6 timings show an 11.782% p95 regression.", "final_result": "Partially fixed. Optional 20-second work does not block the two native row observations. All 180 E6 timings show an 11.782% p95 regression.", "current_evidence": "Native checked rows remain readable during both 20-second optional delays. The final full suite binds provider deadlines, shared budgets and local cancellation tests.", "priority": "P2", "user_impact": "Optional labels delayed publication. Slow response bytes could outlive the job budget. A PG cancel acknowledgment could block the client.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "browser/ui/actual-delayed-names/final-v7/final-receipt.json", "browser/ui/actual-delayed-names/final-v7/transport-events.jsonl", "browser/structural-final-v7/structural-verification.json", "tests/v7/affected.xml", "tests/v7/receipt.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Remaining recheck gap repaired and verified on V5.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "20s optional names/freshness/narration delays expose API-readable rows in57.37/68.03/93.38ms. HTTP dribble and blocked PG cancellation checks pass with actual owned transports. Actual structural DOM readiness is210.78ms, measured separately. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application. Separate final browser injections expose the first25-row page in144.74/189.43/161.23ms during20s names/freshness/narration delays. All three real helpers finish afterward; names/freshness have no lookup work in the invoice fixture.", "verification_snapshot": "final-lineage-v5", "references": [ "api/routers/bi_query.py:9622", "api/routers/bi_query.py:9646", "services/llm_executor.py:94", "services/bi_pg_job_transport.py:19" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json", "evaluation/receipts/v5-final-optional-readiness.json", "browser/optional-v5/optional-ready-to-dom-v5.json" ], "limits": "Remote provider inference termination is not observable. PG14+ connection checks were tested on PG16; older versions use the statement deadline." }, "historical_pre_v14_result_review": { "status": "fixed_in_curated_checks", "root_cause": "Required-ready publication followed optional naming. Inactivity/Future timeouts did not stop underlying provider resources. A cancel acknowledgment could block.", "fix": "Publish checked rows before optional presentation. Kill and reap local provider transports. Close canceled PG connections and retain server deadlines.", "current_evidence": "The V7 real affiliate directory lookup returns three actual rows after a declared 20,005.788 ms transport delay. Required telephone, source authorization, membership and email masking complete before readiness; checked rows are browser-readable 136 ms afterward and remain readable at15 s. Optional text labels patch without interaction loss. The separate unmodified V7 invoice observer records metric DOM insertion 384.559 ms after readiness (submission\u2192metric1336 ms). Earlier no-op name/freshness injections and worker-only timings remain historical, not this final actual-lookup proof. Final V7 affected tests also pass the actual dribbling HTTP response, stalled DNS/TLS and PostgreSQL handshake/cancel/disconnect controls. The existing source admission and terminal-write protections pass. Separate worker checks hold 20-second optional names, freshness and narration below 500ms for required API readiness.", "regression_tests": [ "test_bic015_async_turn_jobs.py", "test_insights_provider_deadline.py", "test_insights_pg_deadline.py", "test_bi_ssh_tunnel_manager.py" ], "limits": "The actual combined browser delay covers affiliate names. Other optional-stage browser injections use the declared invoice fixture; worker controls cover separate delays. Local transport deadlines and reaping are verified. Remote provider inference termination and multi-process capacity remain unverified. One browser observation is not a p95 or a feature speedup ratio.", "final_result": "fixed_in_curated_checks" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_insights_shared_presentation_budget_review.py", "test_insights_v8_provider_attempt_deadline.py", "test_insights_pg_deadline.py" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "V14 listing and total checked DOM appear after 247.572 and 192.163 ms during actual 20,003.060 and 20,000.525 ms optional delays. Parent and optional workers share the six-call and one-retry budget.", "final_result": "V15_frozen_curated_evidence; independent_quality_targets_unreported", "verification_snapshot": "final-contract-v7", "limits": "Layout plus two animation frames is not pixel paint or p95. No remote inference termination or multiprocess capacity proof. E6 remains separate.", "root_cause": "Required-ready publication followed optional naming. Inactivity/Future timeouts did not stop underlying provider resources. A cancel acknowledgment could block.", "fix": "Publish checked rows before optional presentation. Kill and reap local provider transports. Close canceled PG connections and retain server deadlines." }, "v16_independent_result_disposition": { "id": "R07", "title": "Bound optional work and actual local resources", "status": "fixed_on_curated_controls", "root_cause": "Required-ready publication followed optional naming. Inactivity/Future timeouts did not stop underlying provider resources. A cancel acknowledgment could block.", "fix": "Publish checked rows before optional presentation. Kill and reap local provider transports. Close canceled PG connections and retain server deadlines.", "current_evidence": "V16 checked DOM appears 285.054 and 289.813 ms after native readiness while actual optional delays last 20000.384 and 20000.344 ms. Shared six-call and one-retry controls cover optional children. Native jobs all finish.", "regression_tests": [ "test_insights_shared_presentation_budget_review.py", "test_insights_v8_provider_attempt_deadline.py", "test_insights_pg_deadline.py" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json" ], "remaining_limits": "Layout plus two animation frames is not pixel paint or p95. No remote inference termination or multiprocess capacity proof. E6 remains separate.", "final_result": "V16 frozen curated evidence; independent quality targets remain evaluator-owned" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V16 checked DOM appears 285.054 and 289.813 ms after native readiness while actual optional delays last 20000.384 and 20000.344 ms. Shared six-call and one-retry controls cover optional children. Native jobs all finish.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Layout plus two animation frames is not pixel paint or p95. No remote inference termination or multiprocess capacity proof. E6 remains separate.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "R08", "title": "Keep the requested calendar", "status": "fixed", "problem": "Calendar buckets and windows could shift membership under the actual London session, including authored double conversion of timestamptz and comparisons mixing wall clocks with aware bounds.", "root_cause": "The earlier rule assumed every CT timestamp was naive UTC, did not consistently preserve an explicit requested zone through postpasses, and could leave incompatible timestamp operands/anchors. SQL text alone cannot establish a physical timestamp storage type.", "fix": "Use the same authorized current CT connection schema types and requested question in generation/repair and final validation. Naive UTC columns become an instant then the selected wall clock; timestamptz becomes that wall clock directly. Normalize compatible bounds and preserve absolute instant ordering. Refuse unsupported/missing typed lineage and meaning-changing approved SQL rather than rewriting an exact approved statement.", "regression_tests": [ "test_insights_typed_calendar_guard.py", "test_insights_query_correctness.py", "test_insights_postgres_values.py" ], "references": [ "bi/london_calendar_guard.py:56", "bi/london_calendar_guard.py:103", "bi/london_calendar_guard.py:213", "bi/london_calendar_guard.py:400", "bi/london_calendar_guard.py:407", "bi/single_shot_post_pass.py:308", "api/routers/bi_query.py:1768", "api/routers/bi_query.py:2431", "services/bi_query_execution_service.py:3764" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Arbitrary SQL calendar expressions remain outside the finite proof. Final limit: Fixed within curated calendar controls. These finite controls do not establish all live date interpretations.", "final_result": "Fixed within curated calendar controls. These finite controls do not establish all live date interpretations.", "current_evidence": "Typed calendar guards, date parameters, calendar operands and real PostgreSQL calendar controls pass on frozen V18 within the 568 critical checks.", "priority": "P2", "user_impact": "UTC bucket SQL changed meaning under the actual Europe/London session.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "tests/v7/receipt.json", "review/v7-fixture-cleanup/v7-broader-classification.json", "query-v8/v13-independent-cross-contract-review.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "V5 calendar counterexamples remain failed historical evidence. Post-V5 typed helper, full-chain propagation and final executable validation are now covered by V7 integrated checks; no new model-quality outcome is claimed.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "partially_fixed", "current_evidence": "V5 curated cases pass. Four live-model observations show timestamp-type/requested-zone mismatches. Static naive CT timestamp assumptions were applied to an authorized timestamptz source. Post-V5 typed calendar repair is in progress. Post-V5: the typed calendar helper passes 166 tests, including 55 new checks and real London-session PostgreSQL naive/aware, UTC override and DST cases. The final executable SQL hook now uses the authorized current CT schema. Final integrated evaluation remains pending.", "verification_snapshot": "final-lineage-v5", "references": [ "bi/london_calendar_guard.py:120" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json" ], "limits": "UTC, London, New York, DST, null and boundary fixtures pass. Customer connection settings were not inspected." }, "source_reference_evidence": [ { "reference": "src/institutional_kb/bi/london_calendar_guard.py:56", "relevance": "Recognize requested calendar zone without silently choosing between conflicting zones.", "line_text": "def _requested_timezone(query: str) -> str | None:", "sha256": "c4a591c937b8d054b6a3b56d330963628acbdcfa5b36385f7b5f08d2d85e6fb7", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/bi/london_calendar_guard.py:103", "relevance": "Distinguish physical naive/aware/date storage types.", "line_text": "def _storage_kind(data_type: str) -> str | None:", "sha256": "c4a591c937b8d054b6a3b56d330963628acbdcfa5b36385f7b5f08d2d85e6fb7", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/bi/london_calendar_guard.py:213", "relevance": "Normalize supported operands/anchors using physical source lineage and authorized types.", "line_text": "def _normalize_calendar(", "sha256": "c4a591c937b8d054b6a3b56d330963628acbdcfa5b36385f7b5f08d2d85e6fb7", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/bi/london_calendar_guard.py:400", "relevance": "Generated SQL repair contract.", "line_text": "def enforce_london_calendar_buckets(", "sha256": "c4a591c937b8d054b6a3b56d330963628acbdcfa5b36385f7b5f08d2d85e6fb7", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/bi/london_calendar_guard.py:407", "relevance": "Approved SQL validation refuses meaning-changing repair without mutation.", "line_text": "def validate_calendar_buckets(", "sha256": "c4a591c937b8d054b6a3b56d330963628acbdcfa5b36385f7b5f08d2d85e6fb7", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/bi/single_shot_post_pass.py:308", "relevance": "Single-shot postpass forwards question and physical type map.", "line_text": "enforced = enforce_london_calendar_buckets(enforced, query=query, column_types=column_types)", "sha256": "b6f93f0c293c69652687e945f503acf47e4077c86c99e47e558b20cd9670add2", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/api/routers/bi_query.py:1768", "relevance": "Shared CT postpass carries requested question and type map.", "line_text": "enforced = enforce_london_calendar_buckets(", "sha256": "afc564ebf7aa73c48e730e72479eaca4bf22d2f813e4be7edffd5fbe655da2c2", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/api/routers/bi_query.py:2431", "relevance": "CT generation hook obtains authorized CT schema types.", "line_text": "ct_post_process = lambda generated: _enforce_common_ct_post_passes( # noqa: E731", "sha256": "afc564ebf7aa73c48e730e72479eaca4bf22d2f813e4be7edffd5fbe655da2c2", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/services/bi_query_execution_service.py:3764", "relevance": "Final validation checks actual executable SQL and CT server-owned source affinity.", "line_text": "def _validate_final_sql(", "sha256": "94c56c88344233f30748ac7871da6fb8d67b1ca9780f1234607762b568d04958", "matches_frozen_v7": true } ], "test_evidence": [ { "file": "tests/test_insights_typed_calendar_guard.py", "sha256": "76822a3fe7211dcfd1fc89e4be13dd3247092cdc18fb4c6783bacdef50fb22b5", "v7_affected_counts": { "passed": 55 }, "exact_test_references": [ "tests/test_insights_typed_calendar_guard.py:67 (test_physical_storage_and_requested_zone_determine_members)", "tests/test_insights_typed_calendar_guard.py:94 (test_timestamptz_anchors_do_not_coerce_wall_clocks_through_london)", "tests/test_insights_typed_calendar_guard.py:139 (test_single_shot_chain_carries_authorized_types_and_original_requested_zone)", "tests/test_insights_typed_calendar_guard.py:158 (test_authoritative_missing_or_different_quoted_identity_refuses)", "tests/test_insights_typed_calendar_guard.py:179 (test_approved_incorrect_double_conversion_requires_repair_without_mutation)", "tests/test_insights_typed_calendar_guard.py:189 (test_absolute_instant_window_keeps_repeated_dst_hour_order)" ] }, { "file": "tests/test_insights_query_correctness.py", "sha256": "438c2614733c88189c3faf43de913d9f7293ca7946292b927f824d0d8eced802", "v7_affected_counts": { "passed": 96 }, "exact_test_references": [ "tests/test_insights_query_correctness.py:661 (test_approved_calendar_uses_actual_ct_type_without_rewriting)" ] }, { "file": "tests/test_insights_postgres_values.py", "sha256": "acc1a1682f892cc104a44e5831a927f7163bbba271485d707f5534c54ac250e5", "v7_affected_counts": { "passed": 22 }, "exact_test_references": [ "tests/test_insights_postgres_values.py:88 (test_explicit_utc_request_keeps_utc_month_membership)", "tests/test_insights_postgres_values.py:76 (test_london_dst_hour_membership)" ] } ], "historical_pre_v13": { "status": "fixed_in_curated_checks", "current_evidence": "V7 passes all 55 typed-calendar cases with no skips, including actual PostgreSQL execution through the application London connection helper. Independent member sets distinguish UTC, London and New York for naive and aware storage; authored double conversion, aware anchors, DST repeated-hour ordering, nulls, positional/quoted aliases, idempotence, approved-correct SQL, unsupported three-argument DATE_TRUNC and authoritative missing metadata controls pass. V7 final affected execution: 2,234 passed in 74 files, no failures or skips. Broader: 3,704 passed, 10 failures matching the retained original baseline, 13 skipped in 184 files. Mypy: 1,166 source files clean; Ruff clean. These are structural/curated checks; no V7 model-answer outcome is asserted here.", "final_result": "fixed_in_curated_checks", "verification_snapshot": "final-contract-v7" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V15 fullsuite exposes a rejected typed date parameter. V16 finite repair has4redfailures and125green calendar/neighbors, including realPostgreSQL. Independent13controls pass.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "FinalV16 sourcebinding and fullsuite are pending; arbitrary SQL calendar expressions remain outside finite proof.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "R09", "title": "Commit schema and ledger together", "status": "fixed", "problem": "A crash after bic_050 schema commit but before its ledger row made restart fail.", "root_cause": "A transaction-bearing migration could commit schema changes before the runner inserted its ledger row, so restart repeated non-idempotent DDL.", "fix": "For supported explicitly transactional scripts, insert the migration ledger immediately before their COMMIT in the same transaction; preserve trigger bodies and PRAGMA boundaries. Reject unsupported transaction layouts. Leave ordinary scripts on their existing runner path.", "regression_tests": [ "test_insights_migration_recovery.py", "test_insights_result_source_provenance.py" ], "references": [ "db.py:267", "db.py:351" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "No universal claim covers every production database history. Final limit: Fixed in nine final migration recovery controls, including injected pre-ledger and post-commit crashes.", "final_result": "Fixed in nine final migration recovery controls, including injected pre-ledger and post-commit crashes.", "current_evidence": "All nine migration recovery controls pass on V18. Schema and ledger remain atomic across injected failure and restart. Native migrated state persists terminal classes and suggestions.", "priority": "P1", "user_impact": "A crash after bic_050 schema commit but before its ledger row made restart fail.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "tests/v7/receipt.json", "review/v7-fixture-cleanup/v7-broader-classification.json", "evaluation/receipts/r09-final-all-266-migrations.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "The original hard-death repair is unchanged in V7; seven regression cases were rerun by the final affected runner. Historical all-migration evidence is reused only after exact db.py and 266-file hash comparison.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "Hard-death commit/ledger tests and repeated startup pass. All266 current migrations apply and repeat with zero foreign-key violations. Old rows and request keys are preserved. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "db.py:267", "db.py:352" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json", "evaluation/receipts/r09-final-all-266-migrations.json" ], "limits": "All 266 current migrations apply twice. Hard-death tests preserve old rows and keys. Unknown old schemas are not marked applied." }, "source_reference_evidence": [ { "reference": "src/institutional_kb/db.py:267", "relevance": "Main migration runner delegates supported transactional scripts.", "line_text": "def run_migrations(conn: sqlite3.Connection, migrations_dir: Optional[Path] = None) -> None:", "sha256": "eed8d8231ef8491694feb27464d008c8d9c9ce0d1753ec7fd8010f51a03d04bb", "matches_frozen_v7": true }, { "reference": "src/institutional_kb/db.py:351", "relevance": "Validate transaction layout and insert ledger before its COMMIT.", "line_text": "def _run_transaction_migration(conn: sqlite3.Connection, sql: str, name: str) -> bool:", "sha256": "eed8d8231ef8491694feb27464d008c8d9c9ce0d1753ec7fd8010f51a03d04bb", "matches_frozen_v7": true } ], "test_evidence": [ { "file": "tests/test_insights_migration_recovery.py", "sha256": "db7235afd4a245e0415089339b164bc9065de9efb1f2d2e4e558e875309d897e", "v7_affected_counts": { "passed": 7 }, "exact_test_references": [ "tests/test_insights_migration_recovery.py:44 (test_durable_job_migration_restart_is_atomic)", "tests/test_insights_migration_recovery.py:99 (test_transaction_migration_keeps_trigger_bodies_and_pragma_boundaries)", "tests/test_insights_migration_recovery.py:126 (test_process_death_at_migration_boundary_preserves_atomic_recovery)" ] } ], "historical_pre_v14_result_review": { "status": "fixed_in_curated_checks", "root_cause": "A transaction-bearing migration could commit schema changes before the runner inserted its ledger row, so restart repeated non-idempotent DDL.", "fix": "For supported explicitly transactional scripts, insert the migration ledger immediately before their COMMIT in the same transaction; preserve trigger bodies and PRAGMA boundaries. Reject unsupported transaction layouts. Leave ordinary scripts on their existing runner path.", "current_evidence": "V7 passes all seven migration-recovery cases, including process death after rename, before ledger, before commit and after commit; schema/ledger atomicity, repeated restart and legacy request-key preservation are asserted. Historical all-266 startup/restart evidence is reused explicitly: its db.py SHA and every migration SHA still match V7, its second ledger is unchanged and foreign-key violations are zero. V7 final affected execution: 2,234 passed in 74 files, no failures or skips. Broader: 3,704 passed, 10 failures matching the retained original baseline, 13 skipped in 184 files. Mypy: 1,166 source files clean; Ruff clean. These are structural/curated checks; no V7 model-answer outcome is asserted here.", "regression_tests": [ "test_insights_migration_recovery.py" ], "limits": "This closes the demonstrated bic_050 explicit-transaction crash window. It is not a universal atomicity claim for plain scripts, special legacy migration paths, unknown old schemas, or every possible production database history. The all-266 run is retained earlier execution on byte-identical db.py/migrations, not a newly rerun V7 experiment.", "final_result": "fixed_in_curated_checks" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_insights_migration_recovery.py", "test_insights_result_source_provenance.py" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "Root added explicittransaction to bic051 so terminalclass/suggestions columns andmigrationledger commit together. RealSQLite savedturns now roundtrip those fields.", "final_result": "root_owned_recovery_gate; native_V15_terminal_persistence_observed", "verification_snapshot": "final-contract-v7", "limits": "Root owns crash/recovery gate and should attach currentreceipt. No universal olddatabase-history claim.", "root_cause": "A transaction-bearing migration could commit schema changes before the runner inserted its ledger row, so restart repeated non-idempotent DDL.", "fix": "For supported explicitly transactional scripts, insert the migration ledger immediately before their COMMIT in the same transaction; preserve trigger bodies and PRAGMA boundaries. Reject unsupported transaction layouts. Leave ordinary scripts on their existing runner path." }, "v16_independent_result_disposition": { "id": "R09", "title": "Commit schema and ledger together", "status": "fixed_on_curated_controls", "root_cause": "A transaction-bearing migration could commit schema changes before the runner inserted its ledger row, so restart repeated non-idempotent DDL.", "fix": "For supported explicitly transactional scripts, insert the migration ledger immediately before their COMMIT in the same transaction; preserve trigger bodies and PRAGMA boundaries. Reject unsupported transaction layouts. Leave ordinary scripts on their existing runner path.", "current_evidence": "The V16 migrated SQLite database persists and reloads both terminal classes, suggestions and server guard/code provenance. Root owns transactional migration/crash-recovery checks; this native gate proves saved reads only.", "regression_tests": [ "test_insights_migration_recovery.py", "test_insights_result_source_provenance.py" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json" ], "remaining_limits": "Root owns crash/recovery gate and should attach currentreceipt. No universal olddatabase-history claim.", "final_result": "Root-owned migration gate; native V16 terminal persistence observed" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "The V16 migrated SQLite database persists and reloads both terminal classes, suggestions and server guard/code provenance. Root owns transactional migration/crash-recovery checks; this native gate proves saved reads only.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Root owns crash/recovery gate and should attach currentreceipt. No universal olddatabase-history claim.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F01", "title": "Conversation ownership", "status": "fixed", "problem": "Conversation ownership differed across endpoints.", "root_cause": "Endpoints did not consistently apply requesting-user and vault scoping before result access or dispatch. Auxiliary source authorization is a separate boundary.", "fix": "Retain owner/vault checks before dispatch and bind customization to the actual saved source.", "regression_tests": [ "test_insights_auxiliary_authorization.py", "test_insights_customization_source_identity.py", "test_bic007_bi_conversation_api.py:119", "test_bic007_bi_conversation_api.py:258" ], "references": [ "api/routers/bi_query.py:10113", "api/routers/bi_query.py:10929" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Finite authorization tests pass. Production identity-provider integration was not deployed. Final limit: Fixed in the curated ownership and authorization controls. No claim of exhaustive security proof.", "final_result": "Fixed in the curated ownership and authorization controls. No claim of exhaustive security proof.", "current_evidence": "Conversation ownership and auxiliary authorization pass in the unchanged full-suite gate. Fresh native browser calls use the declared synthetic vault.", "priority": "P1", "user_impact": "Conversation ownership differed across endpoints.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "review/v7-fixture-cleanup/v7-broader-classification.json", "evaluation/receipts/v7-final-check-summary.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Earlier fix retained and reverified; related source-identity or recovery repairs are recorded separately. This is not claimed as wholly new behavior.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "api/routers/bi_query.py:10015", "api/routers/bi_query.py:10831" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json" ], "limits": "Production identity-provider integration was not deployed." }, "status_before_current_final_verification": "fixed_in_curated_checks", "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V7: 2,234 affected Python tests pass (74 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 live skips (184 files). Source hashes match frozen V7.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Production identity-provider integration was not deployed.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F02", "title": "SQL authorization", "status": "fixed", "problem": "Incomplete relation extraction could miss restricted tables.", "root_cause": "Relation and column extraction did not resolve qualified physical sources and their policies through lexical SQL scopes.", "fix": "Retain qualified AST validation and policy checks; do not bypass them for templates, repairs or follow-ups.", "regression_tests": [ "test_insights_query_correctness.py", "test_insights_auxiliary_authorization.py" ], "references": [ "bi/sql_source_validation.py:51", "bi/sql_validator.py:1189", "bi/sql_validator.py:1343", "bi/sql_validator.py:1400" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "The repaired read-only fixture is verified. No customer permission configuration was queried. Final limit: Fixed in curated SQL authorization controls. Denials occur before dispatch.", "final_result": "Fixed in curated SQL authorization controls. Denials occur before dispatch.", "current_evidence": "All 568 curated critical controls pass, including SQL allowlist, source identity, masking, and authorization-before-dispatch tests.", "priority": "P1", "user_impact": "Incomplete relation extraction could miss restricted tables.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "review/v7-fixture-cleanup/v7-broader-classification.json", "evaluation/receipts/v7-final-check-summary.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Earlier fix retained and reverified; related source-identity or recovery repairs are recorded separately. This is not claimed as wholly new behavior.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "bi/sql_source_validation.py:51", "bi/sql_validator.py:1189", "bi/sql_validator.py:1343", "bi/sql_validator.py:1400" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json" ], "limits": "Read-only fixture roles passed. No customer policies were queried." }, "status_before_current_final_verification": "fixed_in_curated_checks", "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V7: 2,234 affected Python tests pass (74 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 live skips (184 files). Source hashes match frozen V7.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Read-only fixture roles passed. No customer policies were queried.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F03", "title": "Merge meaning", "status": "partially_fixed", "problem": "Recovery could change membership or measure grain.", "root_cause": "Fallback dropped accepted membership obligations, while aggregation lacked proved physical row identity and join cardinality. See R01 and R02.", "fix": "Apply R01 and R02 through fallback, rescue and aggregation.", "regression_tests": [ "test_churn_tier_routing_and_merge_rescue.py", "test_prd004_federated_merge.py" ], "references": [ "services/bi_federated_execution_service.py:2227", "services/bi_federated_merge.py:903" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Supported route proofs remain finite. A successful merge does not prove the whole question. Final limit: Partially fixed. The held run still has wrong source membership and scalar requests that become per-client listings.", "final_result": "Partially fixed. The held run still has wrong source membership and scalar requests that become per-client listings.", "current_evidence": "V18 query and merge gates pass. Native three-turn scoped aggregation is correct. The held run is exposing wrong source routing and incorrect count/listing shapes.", "priority": "P1", "user_impact": "Recovery could change membership or measure grain.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "review/v7-fixture-cleanup/v7-broader-classification.json", "evaluation/receipts/v7-final-check-summary.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Original finding carried forward; relevant controls retained and challenged. Related R findings describe repairs made in this goal.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "services/bi_federated_execution_service.py:2186", "services/bi_federated_merge.py:903" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json" ], "limits": "Unnecessary refusal is a separate live-evaluation outcome." }, "status_before_current_final_verification": "fixed_in_curated_checks", "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V7: 2,234 affected Python tests pass (74 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 live skips (184 files). Source hashes match frozen V7.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Unnecessary refusal is a separate live-evaluation outcome.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F04", "title": "Displayed values", "status": "partially_fixed", "problem": "Totals could disagree about row limits, units or ID types.", "root_cause": "Aggregation and serializers lost native unit lineage. Some views inferred measures from aliases before honoring authoritative types. See R05.", "fix": "Apply R05 and retain the 499/500/501/600-row, null, negative, zero/one, CSV and share checks.", "regression_tests": [ "test_insights_count_metadata.py", "js/bi-answer-values.test.mjs", "test_insights_v17_explicit_output.py", "test_insights_v17_output_recovery_native.py" ], "references": [ "static/js/bi.js:2783", "static/js/bi.js:3544", "static/js/bi.js:10069", "static/js/bi.js:10102", "static/js/bi.js:13464", "bi/answer_record.py:10" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "The fixed latency suite still returns a wrong settled-revenue field alias on three repeats. General model output correctness remains open. Final limit: Partially fixed. The six native cases agree on checked values, but 54/250 held answers are wrong and 21 comparisons are unverified.", "final_result": "Partially fixed. The six native cases agree on checked values, but 54/250 held answers are wrong and 21 comparisons are unverified.", "current_evidence": "V18 native table, headline, snapshot CSV, share and reload agree on 45 clients, GBP 2454.50, and count 45. Explicit projection and ordering pass native PostgreSQL controls.", "priority": "P1", "user_impact": "Totals could disagree about row limits, units or ID types.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "browser/ui/terminal-ready-recovery/receipt.json", "browser/structural-final-v7/structural-verification.json", "browser/structural-final-v7/snapshot.csv", "tests/v7/affected.xml", "tests-v17/scoped/receipt.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Original finding carried forward; relevant controls retained and challenged. Related R findings describe repairs made in this goal.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "partially_fixed", "current_evidence": "V5 curated cases pass. Opened live evaluation finds native pipeline_value missing unit lineage, and proven identifier aggregation overwritten by numeric metadata. Post-V5 repair is in progress. Post-V5: native pipeline_value unit and identifier repairs pass 229 scoped Python, 102 separate temporary-PostgreSQL, 23 Node value and four PG-to-JS helper controls. Final integrated evaluation remains pending.", "verification_snapshot": "final-lineage-v5", "references": [ "static/js/bi.js:2783", "static/js/bi.js:10070", "static/js/bi.js:13453", "bi/answer_record.py:10" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json", "review/ui-final-challenge.md" ], "limits": "Curated and one homogeneous actual structural fixture pass. Live oracle coverage omits units for some tasks; broad surface correctness needs separate evidence. An already-collapsed unknown-currency scalar cannot recover its units from display metadata. Pre-execution validation is required." }, "historical_pre_v14_result_review": { "status": "fixed_in_curated_checks", "root_cause": "Aggregation and serializers lost native unit lineage. Some views inferred measures from aliases before honoring authoritative types. See R05.", "fix": "Apply R05 and retain the 499/500/501/600-row, null, negative, zero/one, CSV and share checks.", "current_evidence": "The exact V7 JavaScript is covered by 43 Node checks and 56 Chromium cases in two commands (55+1), retaining row-boundary499/500/501/600,0/1, null/negative, partial, nonadditive, percentage, exact integer, native currency and share/CSV cases. Actual V7 preserves50 GBP 24 rows andGBP 1,200 total; the chart uses invoice_id categories and one money series.", "regression_tests": [ "js/bi-answer-values.test.mjs", "e2e/insights-review-fixes.spec.js", "test_insights_answer_record.py", "test_insights_quoted_measure_identity.py", "test_insights_quoted_currency_pins.py" ], "limits": "Curated mocked-API adversaries plus one homogeneous actual money fixture. No arbitrary-query, all-native-source, human-usefulness or live-model correctness claim; pre-execution metadata authority remains a backend dependency.", "final_result": "fixed_in_curated_checks" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_insights_count_metadata.py", "js/bi-answer-values.test.mjs" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "V15 native 45-row listing, GBP 2454.50 total and 45 count agree on all checked saved surfaces. Current 244 Node and 56 mocked Chromium tests cover units, nulls, sorting, CSV/share and row limits.", "final_result": "V15_frozen_curated_evidence; independent_quality_targets_unreported", "verification_snapshot": "final-contract-v7", "limits": "No mixed-currency native-browser or general model precision claim. Refusal records now agree; transient footer wording remains open.", "root_cause": "Aggregation and serializers lost native unit lineage. Some views inferred measures from aliases before honoring authoritative types. See R05.", "fix": "Apply R05 and retain the 499/500/501/600-row, null, negative, zero/one, CSV and share checks." }, "v16_independent_result_disposition": { "id": "F04", "title": "Displayed values", "status": "partially_fixed", "root_cause": "Aggregation and serializers lost native unit lineage. Some views inferred measures from aliases before honoring authoritative types. See R05.", "fix": "Apply R05 and retain the 499/500/501/600-row, null, negative, zero/one, CSV and share checks.", "current_evidence": "V16 native 45-client listing, GBP 2454.50 total and scalar count45 agree on all checked saved surfaces. Fresh 244 Node, 56 main and 21 terminal mocked Chromium controls include values, nulls, sorting, exports and source labels. The neutral ready footer removes the retained V15 representation defect.", "regression_tests": [ "test_insights_count_metadata.py", "js/bi-answer-values.test.mjs" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json", "results-v8/v17-rank-threshold-prepared/rank-threshold-red.json" ], "remaining_limits": "No mixed-currency native-browser or general model precision claim. Five zero/row/error/refusal/clarification ready states use mocked Chromium. V16 development also proves an open display-cap regression: at least one settled booking serves only one of seven checked clients.", "final_result": "V16 native curated gate passed; later development wrong-success blocker remains open" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V16 native saved surfaces agree on checked values. Two development successes still return wrong requested fields or display order. V17 explicit output and bounded-retry controls are in progress. V17 final scoped gate passes 701 checks with no skips; native bounded correction returns both retained output cases correctly. Frozen full/native/live verification is pending.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "No mixed-currency native-browser or general model precision claim. Refusal records now agree; transient footer wording remains open.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F05", "title": "Semantic proof", "status": "partially_fixed", "problem": "Positional outputs could rename category to status. PostgreSQL quoted Status or Clients could be confused with the required lowercase physical identifier.", "root_cause": "The verifier lost positional lineage and folded case-sensitive physical identifiers. Remaining cause: Generated SQL can coincide with one fixture value while using a different predicate or entity key.", "fix": "Trace positional outputs. Normalize only PostgreSQL unquoted names. Resolve aliases before checking physical identity. Unsupported quoted physical names return unverified.", "regression_tests": [ "test_insights_query_correctness.py", "test_insights_postgres_values.py", "test_insights_quoted_semantic_identity.py", "test_insights_v17_explicit_output.py" ], "references": [ "bi/semantic_obligations.py:51", "bi/semantic_obligations.py:70", "bi/semantic_obligations.py:94" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Frozen numeric scores remain unchanged. Finite fixture equality is not complete semantic proof. Final limit: Partially fixed; general semantic proof is disproved. Two scorer-correct first turns fail independent SQL counterexamples. Eligibility, earliest-date and identity defects remain.", "final_result": "Partially fixed; general semantic proof is disproved. Two scorer-correct first turns fail independent SQL counterexamples. Eligibility, earliest-date and identity defects remain.", "current_evidence": "The 568 critical controls pass. Two held first-turn answers that score correct fail read-only synthetic semantic counterexamples: NULL role is treated as User; note person is substituted for deal person. Final held evidence: Independent native VALUES checks show NULL role treated as User (0 becomes 1), and note person used for deal person (2 becomes 1). Frozen labels stay unchanged.", "priority": "P2", "user_impact": "Positional outputs could rename category to status. PostgreSQL quoted Status or Clients could be confused with the required lowercase physical identifier.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "review/v7-fixture-cleanup/v7-broader-classification.json", "evaluation/receipts/v7-final-check-summary.json", "tests-v17/scoped/receipt.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json", "phase0/final-v18-late-predicate-fidelity-audit.json", "phase0/held-v18-lead-semantic-counterexamples.json" ], "resolution_provenance": "Original finding carried forward; relevant controls retained and challenged. Related R findings describe repairs made in this goal.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "bi/semantic_obligations.py:51", "bi/semantic_obligations.py:70", "bi/semantic_obligations.py:94" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json", "review/final-query-review-v5.md" ], "limits": "36 focused checks include 17 actual PostgreSQL cases. Safe lowercase quotes, unquoted case folding and quoted derived aliases remain supported. The verifier cannot prove arbitrary natural-language meaning." }, "status_before_current_final_verification": "fixed_in_curated_checks", "historical_before_v17_lead_update": { "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V7: 2,234 affected Python tests pass (74 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 live skips (184 files). Source hashes match frozen V7.", "final_result": "fixed_in_curated_checks", "verification_snapshot": "final-contract-v7", "limits": "The current quoted-identity module has 31 passing cases, including 13 actual PostgreSQL cases. Safe lowercase quotes, unquoted case folding and quoted derived aliases remain supported. The verifier does not prove arbitrary natural-language meaning.", "root_cause": "The verifier lost positional lineage and folded case-sensitive physical identifiers.", "fix": "Trace positional outputs. Normalize only PostgreSQL unquoted names. Resolve aliases before checking physical identity. Unsupported quoted physical names return unverified." }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V16 successes can omit requested facts. V17 checks a literal output list and final SQL order, and distinguishes a ranking measure from an identity-only output. General expression correctness remains unproved. V17 final scoped gate passes 701 checks with no skips; native bounded correction returns both retained output cases correctly. Frozen full/native/live verification is pending.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "The current quoted-identity module has 31 passing cases, including 13 actual PostgreSQL cases. Safe lowercase quotes, unquoted case folding and quoted derived aliases remain supported. The verifier does not prove arbitrary natural-language meaning.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F06", "title": "Job and parent protocol", "status": "partially_fixed", "problem": "Retries, completion and follow-up identity could disagree.", "root_cause": "Full-thread context reloads, conflated completion/retrieval states and a separate migration ledger commit allowed inconsistent parent or retry state. See R04, R06 and R09.", "fix": "Apply R04, R06 and R09. Preserve atomic request keys and terminal-write protection.", "regression_tests": [ "test_bic015_async_turn_jobs.py", "test_insights_v8_chain_recovery.py" ], "references": [ "api/routers/bi_query.py:9703", "api/routers/bi_query.py:9741" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Full live conversation success remains a separate required measurement. The native check covers one successful three-turn chain. Final limit: Partially fixed. All 250 jobs finish. Forty-four explicit dependent turns after a failed or refused parent bypass recovery.", "final_result": "Partially fixed. All 250 jobs finish. Forty-four explicit dependent turns after a failed or refused parent bypass recovery.", "current_evidence": "All six native jobs finish. Thirty final reads start no new work. Refusal records agree. The 57 three/five-turn chain controls pass, with failures at each position.", "priority": "P1", "user_impact": "Retries, completion and follow-up identity could disagree.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "review/v7-fixture-cleanup/v7-broader-classification.json", "evaluation/receipts/v7-final-check-summary.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Original finding carried forward; relevant controls retained and challenged. Related R findings describe repairs made in this goal.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "api/routers/bi_query.py:9605", "api/routers/bi_query.py:9643" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json" ], "limits": "Customer multi-process topology is not reproduced." }, "historical_pre_v14_result_review": { "status": "fixed_in_curated_checks", "root_cause": "Full-thread context reloads, conflated completion/retrieval states and a separate migration ledger commit allowed inconsistent parent or retry state. See R04, R06 and R09.", "fix": "Apply R04, R06 and R09. Preserve atomic request keys and terminal-write protection.", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V7: 2,234 affected Python tests pass (74 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 live skips (184 files). Source hashes match frozen V7.", "regression_tests": [ "test_bic015_async_turn_jobs.py", "test_insights_migration_recovery.py" ], "limits": "Customer multi-process topology is not reproduced.", "final_result": "fixed_in_curated_checks" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_bic015_async_turn_jobs.py", "test_insights_v8_chain_recovery.py" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "V15 six native jobs finish with durable top-level outcomes and 30 reads that start no work. Write and grain refusals persist guard/code; blocked-parent recovery names the smallest next step.", "final_result": "V15_frozen_curated_evidence; independent_quality_targets_unreported", "verification_snapshot": "final-contract-v7", "limits": "No customer multiprocess topology is reproduced. The ready lifecycle is correct, but its footer wrongly assumes rows for a refusal.", "root_cause": "Full-thread context reloads, conflated completion/retrieval states and a separate migration ledger commit allowed inconsistent parent or retry state. See R04, R06 and R09.", "fix": "Apply R04, R06 and R09. Preserve atomic request keys and terminal-write protection." }, "v16_independent_result_disposition": { "id": "F06", "title": "Job and parent protocol", "status": "fixed_on_curated_controls", "root_cause": "Full-thread context reloads, conflated completion/retrieval states and a separate migration ledger commit allowed inconsistent parent or retry state. See R04, R06 and R09.", "fix": "Apply R04, R06 and R09. Preserve atomic request keys and terminal-write protection.", "current_evidence": "All six V16 native jobs finish. Thirty final reads start no work. Write and grain refusals persist server guard/code and matching nested outcomes. The failed-parent follow-up asks to rerun or name the group. Five mocked ready-outcome controls prove neutral footer wording.", "regression_tests": [ "test_bic015_async_turn_jobs.py", "test_insights_v8_chain_recovery.py" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json" ], "remaining_limits": "No customer multiprocess topology is reproduced. Native tests cover one synthetic user and two source connections.", "final_result": "V16 frozen curated evidence; independent quality targets remain evaluator-owned" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "All six V16 native jobs finish. Thirty final reads start no work. Write and grain refusals persist server guard/code and matching nested outcomes. The failed-parent follow-up asks to rerun or name the group. Five mocked ready-outcome controls prove neutral footer wording.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "No customer multiprocess topology is reproduced. The ready lifecycle is correct, but its footer wrongly assumes rows for a refusal.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F07", "title": "Early useful rows", "status": "fixed", "problem": "Optional labels and prose blocked checked results.", "root_cause": "Required-ready persistence followed optional naming and prose. A caller waiting on a Future did not stop underlying transport work. See R07.", "fix": "Apply R07. Persist ready rows before optional work and expose separate follow-up readiness.", "regression_tests": [ "test_bic015_async_turn_jobs.py", "test_insights_shared_presentation_budget_review.py" ], "references": [ "api/routers/bi_query.py:9741", "api/routers/bi_query.py:9746", "static/js/bi.js:7428", "static/js/bi.js:7589" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "These are two registered native observations. They are not p95 or pixel-paint timing. E6 fails its separate p95 target. Final limit: Fixed within the measured native controls. Checked rows remain readable in 225.498 ms and 334.198 ms during actual optional delays of at least 20 seconds.", "final_result": "Fixed within the measured native controls. Checked rows remain readable in 225.498 ms and 334.198 ms during actual optional delays of at least 20 seconds.", "current_evidence": "Checked rows are readable 225.498 ms and 334.198 ms after native readiness during optional delays of at least 20 seconds.", "priority": "P1", "user_impact": "Optional labels and prose blocked checked results.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "browser/ui/actual-delayed-names/final-v7/final-receipt.json", "browser/ui/terminal-ready-recovery/receipt.json", "tests/v7/affected.xml", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Original finding carried forward; relevant controls retained and challenged. Related R findings describe repairs made in this goal.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "api/routers/bi_query.py:9622", "api/routers/bi_query.py:9646" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json", "evaluation/receipts/v5-final-optional-readiness.json", "browser/optional-v5/optional-ready-to-dom-v5.json" ], "limits": "Final worker measurements are local fixtures, not an overall speedup ratio." }, "v13_native_evidence": "Correct native45-client listing,2454.50total and45count;30final reads dispatch no execution. Blocked parent offers minimal recovery.", "historical_pre_v14_result_review": { "status": "fixed_in_curated_checks", "root_cause": "Required-ready persistence followed optional naming and prose. A caller waiting on a Future did not stop underlying transport work. See R07.", "fix": "Apply R07. Persist ready rows before optional work and expose separate follow-up readiness.", "current_evidence": "Final V7 actual affiliate lookup delay20,005.788 ms leaves required checked rows readable136 ms after persisted readiness. Follow-up-ready and submit-enabled states are observed before optional completion; required phones, source authorization, membership and masking precede readiness. Eventual names and reload preserve checked values and reading state.", "regression_tests": [ "test_bic015_async_turn_jobs.py" ], "limits": "One actual affiliate case, not a production latency distribution. A concurrent follow-up is not submitted in this actual harness; the same-source JavaScript regression file covers that interaction with mocked APIs. Transport cancellation and other optional stages have separate evidence.", "final_result": "fixed_in_curated_checks" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_bic015_async_turn_jobs.py", "test_insights_shared_presentation_budget_review.py" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "V14 listing and total checked DOM appear within 500 ms of native readiness during actual 20-second optional work. Required and optional times are recorded separately.", "final_result": "V15_frozen_curated_evidence; independent_quality_targets_unreported", "verification_snapshot": "final-contract-v7", "limits": "Two observations under concurrent load are not p95 or a speed ratio. Follow-up during optional work has separate deterministic controls.", "root_cause": "Required-ready persistence followed optional naming and prose. A caller waiting on a Future did not stop underlying transport work. See R07.", "fix": "Apply R07. Persist ready rows before optional work and expose separate follow-up readiness." }, "v16_independent_result_disposition": { "id": "F07", "title": "Early useful rows", "status": "fixed_on_curated_controls", "root_cause": "Required-ready persistence followed optional naming and prose. A caller waiting on a Future did not stop underlying transport work. See R07.", "fix": "Apply R07. Persist ready rows before optional work and expose separate follow-up readiness.", "current_evidence": "Two V16 native checked results remain readable within500 ms of readiness during real20-second optional waits. Required and optional times are recorded separately. No p95 or relative latency claim is made.", "regression_tests": [ "test_bic015_async_turn_jobs.py", "test_insights_shared_presentation_budget_review.py" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json" ], "remaining_limits": "Two observations under concurrent load are not p95 or a speed ratio. Follow-up during optional work has separate deterministic controls.", "final_result": "V16 frozen curated evidence; independent quality targets remain evaluator-owned" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "Two V16 native checked results remain readable within500 ms of readiness during real20-second optional waits. Required and optional times are recorded separately. No p95 or relative latency claim is made.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Two observations under concurrent load are not p95 or a speed ratio. Follow-up during optional work has separate deterministic controls.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F08", "title": "Small status reads", "status": "fixed", "problem": "Polling could deserialize every historical row.", "root_cause": "Status authorization and loading could hydrate result history before returning a small response. Metadata-only scoping and selected-turn retrieval now separate these reads.", "fix": "Retain the small status projection and selected-turn retrieval.", "regression_tests": [ "test_bic015_async_turn_jobs.py", "test_insights_answer_record.py" ], "references": [ "api/routers/bi_query.py:10295", "api/routers/bi_query.py:10113", "api/routers/bi_query.py:10197" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "No production large-history payload benchmark was run. Final limit: Fixed in final native controls. Thirty final reads start zero SQL, provider work or jobs.", "final_result": "Fixed in final native controls. Thirty final reads start zero SQL, provider work or jobs.", "current_evidence": "Thirty fresh native final reads create zero queries, provider calls, jobs, or usage rows. Saved refusals and the neutral ready footer agree.", "priority": "P1", "user_impact": "Polling could deserialize every historical row.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "review/v7-fixture-cleanup/v7-broader-classification.json", "evaluation/receipts/v7-final-check-summary.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Earlier fix retained and reverified; related source-identity or recovery repairs are recorded separately. This is not claimed as wholly new behavior.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "Status-read checks forbid full-history hydration. Actual saved-turn reads return the persisted answer without another job or generation call. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "api/routers/bi_query.py:10197", "api/routers/bi_query.py:10015", "api/routers/bi_query.py:10099" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json" ], "limits": "No new production history-size distribution was measured." }, "historical_pre_v14_result_review": { "status": "fixed_in_curated_checks", "root_cause": "Status authorization and loading could hydrate result history before returning a small response. Metadata-only scoping and selected-turn retrieval now separate these reads.", "fix": "Retain the small status projection and selected-turn retrieval.", "current_evidence": "Status-read checks forbid full-history hydration. Actual saved-turn reads return the persisted answer without another job or generation call. Final V7: 2,234 affected Python tests pass (74 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 live skips (184 files). Source hashes match frozen V7.", "regression_tests": [ "test_bic015_async_turn_jobs.py", "test_insights_answer_record.py" ], "limits": "No new production history-size distribution was measured.", "final_result": "fixed_in_curated_checks" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_bic015_async_turn_jobs.py", "test_insights_answer_record.py" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "V14 final selected-turn reads use saved rows and start no execution. Small status projection remains unchanged.", "final_result": "V15_frozen_curated_evidence; independent_quality_targets_unreported", "verification_snapshot": "final-contract-v7", "limits": "No new large-history distribution or production payload benchmark. Nested answer records now agree. The temporary ready footer still uses row wording.", "root_cause": "Status authorization and loading could hydrate result history before returning a small response. Metadata-only scoping and selected-turn retrieval now separate these reads.", "fix": "Retain the small status projection and selected-turn retrieval." }, "v16_independent_result_disposition": { "id": "F08", "title": "Small status reads", "status": "fixed_on_curated_controls", "root_cause": "Status authorization and loading could hydrate result history before returning a small response. Metadata-only scoping and selected-turn retrieval now separate these reads.", "fix": "Retain the small status projection and selected-turn retrieval.", "current_evidence": "V16 selected-turn final reads use saved rows and cause zero new execution. Both saved refusals retain consistent computed answer records. The temporary ready footer now describes the saved answer without implying successful rows.", "regression_tests": [ "test_bic015_async_turn_jobs.py", "test_insights_answer_record.py" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json" ], "remaining_limits": "No new large-history or production payload benchmark. Native evidence proves saved read behavior on these six turns.", "final_result": "V16 frozen curated evidence; independent quality targets remain evaluator-owned" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V16 selected-turn final reads use saved rows and cause zero new execution. Both saved refusals retain consistent computed answer records. The temporary ready footer now describes the saved answer without implying successful rows.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "No new large-history distribution or production payload benchmark. Nested answer records now agree. The temporary ready footer still uses row wording.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F09", "title": "Freshness and provenance", "status": "partially_fixed", "problem": "Run time, source age and snapshot identity could be conflated.", "root_cause": "Run time, source age and snapshot identity could be conflated.", "fix": "Keep source and snapshot metadata separate. Preserve unknown freshness and actual connection identity.", "regression_tests": [ "test_insights_result_source_provenance.py", "js/bi-federation-source.test.mjs", "js/bi-rowcap-gate.test.mjs" ], "references": [ "bi/answer_record.py:48", "static/js/bi.js:12147", "static/js/bi.js:12316", "api/routers/bi_query.py:13304" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Source watermarks, refreshed CSV after source changes, and atomic cross-source snapshots are not verified. Final limit: Partially fixed. Stored scope and provenance agree in native checks. Generated source predicates and entity identity remain fallible.", "final_result": "Partially fixed. Stored scope and provenance agree in native checks. Generated source predicates and entity identity remain fallible.", "current_evidence": "The native stored snapshot retains checked GBP, NULL and source metadata through export, share and reload. All 397 UI verification inputs match the retained V16 bytes.", "priority": "P2", "user_impact": "Run time, source age and snapshot identity could be conflated.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "browser/structural-final-v7/structural-verification.json", "browser/structural-final-v7/repeated-read-source-proof.json", "browser/ui/actual-delayed-names/final-v7/final-receipt.json", "browser/ui/terminal-ready-recovery/integration-check/receipt.json", "tests/v7/affected.xml", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Original finding carried forward; relevant controls retained and challenged. Related R findings describe repairs made in this goal.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "partially_verified", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "bi/answer_record.py:48", "api/routers/bi_query.py:6668" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json" ], "limits": "No customer watermarks or atomic cross-source snapshot were available." }, "historical_pre_v14_result_review": { "status": "partially_verified", "root_cause": "Run time, source age and snapshot identity could be conflated.", "fix": "Keep source and snapshot metadata separate. Preserve unknown freshness and actual connection identity.", "current_evidence": "V7 rendered receipts visibly distinguish saved snapshot/query time from unknown source refresh and explicitly restrict generation-confidence scope. Invoice saved CSV, share and share reload preserve GBP 1,200; actual-name disclosure remains open during updates. Five selected reads on each actual fixture preserve stored values and execution counters.", "regression_tests": [ "test_insights_answer_record.py", "test_insights_label_source_identity.py" ], "limits": "No customer watermark, atomic cross-source snapshot, real refreshed-CSV source mutation, or live confidence calibration was measured. Exact-rerun refusals are separately inspected adapter/captured-dispatch controls; snapshot success does not certify refreshed execution.", "final_result": "partially_verified" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_insights_result_source_provenance.py", "js/bi-federation-source.test.mjs", "js/bi-rowcap-gate.test.mjs" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "Current Node source tests pass. Physical source_label stays separate from factual measure_class; Gold may host CT snapshot measures. Saved merge replay retains its operational class. V14 query repair excludes explicitly Gold-sourced values from CT scalar reduction.", "final_result": "V15_frozen_curated_evidence; independent_quality_targets_unreported", "verification_snapshot": "final-contract-v7", "limits": "No fresh-data CSV mutation, source watermark or atomic cross-source snapshot proof. Native success cases use one source; merged native case refuses grain.", "root_cause": "Run time, source age and snapshot identity could be conflated.", "fix": "Keep source and snapshot metadata separate. Preserve unknown freshness and actual connection identity." }, "v16_independent_result_disposition": { "id": "F09", "title": "Freshness and provenance", "status": "partially_fixed", "root_cause": "Run time, source age and snapshot identity could be conflated.", "fix": "Keep source and snapshot metadata separate. Preserve unknown freshness and actual connection identity.", "current_evidence": "Fresh V16 Node source tests preserve physical source identity separately from measure basis. The native snapshot listing retains GBP and NULL metadata through export/share/reload. Replay and mixed-source classification remain separately tested backend contracts.", "regression_tests": [ "test_insights_result_source_provenance.py", "js/bi-federation-source.test.mjs", "js/bi-rowcap-gate.test.mjs" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json" ], "remaining_limits": "No fresh-data CSV mutation, source watermark or atomic cross-source snapshot proof. Native success cases use one source; merged native case refuses grain.", "final_result": "V16 frozen curated evidence; independent quality targets remain evaluator-owned" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "Fresh V16 Node source tests preserve physical source identity separately from measure basis. The native snapshot listing retains GBP and NULL metadata through export/share/reload. Replay and mixed-source classification remain separately tested backend contracts.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "No fresh-data CSV mutation, source watermark or atomic cross-source snapshot proof. Native success cases use one source; merged native case refuses grain.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F10", "title": "Empty-result recovery", "status": "partially_fixed", "problem": "Empty views hid SQL inspection and recovery.", "root_cause": "Empty views hid SQL inspection and recovery.", "fix": "Retain SQL access and clear next actions on empty results.", "regression_tests": [ "test_insights_v8_chain_recovery.py", "e2e/insights-review-fixes.spec.js" ], "references": [ "static/js/bi.js:12205", "static/js/bi.js:7493", "static/js/bi.js:8471" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "The six fresh native cases do not include empty successful results or an injected result-read outage. Final limit: Partially fixed. Curated empty-result recovery passes. The held run still gives 24 unverified refusals and wrong membership answers.", "final_result": "Partially fixed. Curated empty-result recovery passes. The held run still gives 24 unverified refusals and wrong membership answers.", "current_evidence": "Empty-result and failed saved-read recovery pass the retained mocked Chromium controls, bound to exact unchanged UI bytes. Fresh native blocked-parent recovery offers rerun or a named group.", "priority": "P2", "user_impact": "Empty views hid SQL inspection and recovery.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "browser/ui/terminal-ready-recovery/receipt.json", "browser/ui/terminal-ready-recovery/full/browser.log", "tests/v7/affected.xml", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Earlier fix retained and reverified; related source-identity or recovery repairs are recorded separately. This is not claimed as wholly new behavior.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "fixed_in_curated_checks", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "static/js/bi.js:12190", "static/js/bi.js:8421" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json", "review/ui-final-challenge.md" ], "limits": "Browser recovery fixtures pass; usefulness has no human score." }, "historical_pre_v14_result_review": { "status": "fixed_in_curated_checks", "root_cause": "Empty views hid SQL inspection and recovery.", "fix": "Retain SQL access and clear next actions on empty results.", "current_evidence": "The retained current-JS Chromium empty/incomplete-result case preserves an explicit limitation and SQL/filter inspection; saved-result retrieval failures retain their own GET retry control rather than substituting a new query.", "regression_tests": [ "e2e/insights-review-fixes.spec.js" ], "limits": "This uses mocked-API browser fixtures; neither actual V7 structural query returns an empty set. No human usefulness score.", "final_result": "fixed_in_curated_checks" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_insights_v8_chain_recovery.py", "e2e/insights-review-fixes.spec.js" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "Current 56-case mocked Chromium suite verifies empty-view SQL and recovery. V14 backend controls prove zero checked members -> one zero count without generation. Native failed-parent recovery asks for a rerun or group.", "final_result": "V15_frozen_curated_evidence; independent_quality_targets_unreported", "verification_snapshot": "final-contract-v7", "limits": "The six native cases contain no empty successful result. Empty-browser observations use mocked API responses.", "root_cause": "Empty views hid SQL inspection and recovery.", "fix": "Retain SQL access and clear next actions on empty results." }, "v16_independent_result_disposition": { "id": "F10", "title": "Empty-result recovery", "status": "fixed_on_curated_controls", "root_cause": "Empty views hid SQL inspection and recovery.", "fix": "Retain SQL access and clear next actions on empty results.", "current_evidence": "Fresh V16 mocked Chromium controls retain empty-view SQL/recovery. Five ready-outcome cases include empty success. Native refused-parent recovery asks for a rerun or group. Native six-case gate does not include an empty successful query.", "regression_tests": [ "test_insights_v8_chain_recovery.py", "e2e/insights-review-fixes.spec.js" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json" ], "remaining_limits": "The six native cases contain no empty successful result. Empty-browser observations use mocked API responses.", "final_result": "V16 frozen curated evidence; independent quality targets remain evaluator-owned" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "Fresh V16 mocked Chromium controls retain empty-view SQL/recovery. Five ready-outcome cases include empty success. Native refused-parent recovery asks for a rerun or group. Native six-case gate does not include an empty successful query.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "The six native cases contain no empty successful result. Empty-browser observations use mocked API responses.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F11", "title": "Accessibility and mobile", "status": "partially_fixed", "problem": "Some controls, focus paths and chart alternatives failed.", "root_cause": "Some controls, focus paths and chart alternatives failed.", "fix": "Preserve focus and expose keyboard, table and mobile recovery paths.", "regression_tests": [ "e2e/insights-review-fixes.spec.js" ], "references": [ "static/js/bi.js:3544", "static/js/bi.js:4105", "static/js/bi.js:7428", "static/js/bi.js:12147", "static/js/bi.js:13464" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "No screen reader, physical touch, or human usability study was run. Final limit: Partially fixed. Retained mocked UI checks are bound to 397 unchanged inputs. Screen readers and physical touch remain unverified.", "final_result": "Partially fixed. Retained mocked UI checks are bound to 397 unchanged inputs. Screen readers and physical touch remain unverified.", "current_evidence": "The 244 Node and 77 mocked Chromium passes are source-bound to unchanged V18 UI inputs. Fresh native desktop screenshots show the total, explicit refusal, and recovery.", "priority": "P2", "user_impact": "Some controls, focus paths and chart alternatives failed.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "browser/ui/terminal-ready-recovery/receipt.json", "browser/structural-final-v7/integrated-desktop.png", "browser/structural-final-v7/integrated-mobile.png", "browser/structural-final-v7/integrated-share.png", "browser/ui/actual-delayed-names/final-v7/final-receipt.json", "tests/v7/affected.xml", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Original finding carried forward; relevant controls retained and challenged. Related R findings describe repairs made in this goal.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "partially_verified", "current_evidence": "Final47 Chromium checks cover keyboard/focus/mobile and table alternatives. The actual invoice chart still repeatsGBP labels. No screen reader was observed. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "static/js/bi.js:7541", "static/js/bi.js:8421", "static/js/bi.js:13453" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json", "review/ui-final-challenge.md" ], "limits": "No screen-reader observation. Actual invoice chart repeats GBP as its dimension and remains weak." }, "historical_pre_v14_result_review": { "status": "partially_verified", "root_cause": "Some controls, focus paths and chart alternatives failed.", "fix": "Preserve focus and expose keyboard, table and mobile recovery paths.", "current_evidence": "Current exact-JS Chromium coverage includes dark/light mobile focus, inert/trapped rails and question library, chart canvas names and exact-data alternative, sorted-column focus preservation, receipt state, and recovery. Actual V7 desktop/mobile/share screenshots inspected. Actual invoice chart now labels invoice_id categories instead of repeated GBP, with identifiers excluded from measures. Actual delayed naming preserves a focused draft and nonzero reading position.", "regression_tests": [ "e2e/insights-review-fixes.spec.js" ], "limits": "No actual screen reader, mobile touch interaction or human usability rating. Actual structural screenshots are dark mode; light-theme interactions are browser fixtures. Invoice plot still has50equal-height bars; first-row telephone/amount headline is factual but weak synthesis. The people-ID display lookup remains unavailable under this fixture masking policy, while actual affiliate partner labels succeed.", "final_result": "partially_verified" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "e2e/insights-review-fixes.spec.js" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "The 56 main and 16 terminal/submission Chromium cases executed on V14 are reused after exact equality of every test/config and all142 static files. V15 dark desktop native screenshots were inspected, including focused total and grain refusal.", "final_result": "V15_frozen_curated_evidence; independent_quality_targets_unreported", "verification_snapshot": "final-contract-v7", "limits": "No screen-reader, physical touch or human usability study. DOM layout is not a pixel-paint measurement.", "root_cause": "Some controls, focus paths and chart alternatives failed.", "fix": "Preserve focus and expose keyboard, table and mobile recovery paths." }, "v16_independent_result_disposition": { "id": "F11", "title": "Accessibility and mobile", "status": "fixed_on_curated_controls", "root_cause": "Some controls, focus paths and chart alternatives failed.", "fix": "Preserve focus and expose keyboard, table and mobile recovery paths.", "current_evidence": "V16 ran fresh 56 main and21 terminal Chromium checks, including two themes and five ready-outcome lifecycle controls. Native dark desktop grain refusal and shared scoped-total screenshots were inspected. No screen-reader or physical-touch claim is made.", "regression_tests": [ "e2e/insights-review-fixes.spec.js" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json" ], "remaining_limits": "No screen-reader, physical touch or human usability study. DOM layout is not a pixel-paint measurement.", "final_result": "V16 frozen curated evidence; independent quality targets remain evaluator-owned" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V16 ran fresh 56 main and21 terminal Chromium checks, including two themes and five ready-outcome lifecycle controls. Native dark desktop grain refusal and shared scoped-total screenshots were inspected. No screen-reader or physical-touch claim is made.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "No screen-reader, physical touch or human usability study. DOM layout is not a pixel-paint measurement.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F12", "title": "Bounded resources", "status": "partially_fixed", "problem": "Connection and model work had incomplete timeout or admission evidence.", "root_cause": "Connection and model work had incomplete timeout or admission evidence.", "fix": "Retain process-wide admission and bound local HTTP, DNS, SSH and PG work.", "regression_tests": [ "test_insights_v8_provider_attempt_deadline.py", "test_insights_shared_presentation_budget_review.py", "test_insights_pg_deadline.py" ], "references": [ "services/llm_executor.py:94", "services/bi_pg_job_transport.py:19" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Remote inference termination and multiprocess capacity are not proved. E6 p95 rises 11.782%, above the 10% limit. Final limit: Partially fixed. Zero of 250 candidate jobs reaches the deadline. Twenty provider-boundary failures remain. Deterministic p95 regresses 11.782%.", "final_result": "Partially fixed. Zero of 250 candidate jobs reaches the deadline. Twenty provider-boundary failures remain. Deterministic p95 regresses 11.782%.", "current_evidence": "Shared generation and retry budgets pass final controls. All six native jobs finish. E6 preserves all 180 attempts and confirms the deterministic adapter.", "priority": "P2", "user_impact": "Connection and model work had incomplete timeout or admission evidence.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "review/v7-fixture-cleanup/v7-broader-classification.json", "evaluation/receipts/v7-final-check-summary.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Original finding carried forward; relevant controls retained and challenged. Related R findings describe repairs made in this goal.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "partially_verified", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "services/llm_executor.py:94", "services/bi_pg_job_transport.py:19" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json", "evaluation/receipts/v5-final-optional-readiness.json" ], "limits": "Multi-process capacity and complete remote provider resource usage remain unverified." }, "historical_pre_v14_result_review": { "status": "partially_verified", "root_cause": "Connection and model work had incomplete timeout or admission evidence.", "fix": "Retain process-wide admission and bound local HTTP, DNS, SSH and PG work.", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V7: 2,234 affected Python tests pass (74 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 live skips (184 files). Source hashes match frozen V7.", "regression_tests": [ "test_insights_provider_deadline.py", "test_insights_pg_deadline.py", "test_bi_host_policy.py" ], "limits": "Multi-process capacity and complete remote provider resource usage remain unverified.", "final_result": "partially_verified" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_insights_v8_provider_attempt_deadline.py", "test_insights_shared_presentation_budget_review.py", "test_insights_pg_deadline.py" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "Absolute provider HTTP deadlines, shared SQL/optional budgets, local process cancellation and PostgreSQL deadline controls remain. V15 native six jobs finish; optional work does not hide checked rows.", "final_result": "V15_frozen_curated_evidence; independent_quality_targets_unreported", "verification_snapshot": "final-contract-v7", "limits": "Remote resource use and multiprocess capacity are not proved. A provider exception alone does not prove a remote outage.", "root_cause": "Connection and model work had incomplete timeout or admission evidence.", "fix": "Retain process-wide admission and bound local HTTP, DNS, SSH and PG work." }, "v16_independent_result_disposition": { "id": "F12", "title": "Bounded resources", "status": "fixed_on_curated_controls", "root_cause": "Connection and model work had incomplete timeout or admission evidence.", "fix": "Retain process-wide admission and bound local HTTP, DNS, SSH and PG work.", "current_evidence": "V16 native six jobs finish; optional work does not block checked rows. Absolute provider HTTP deadlines, concurrent shared budgets and local cancellation have retained focused controls. The empty-Gold concurrent-provider drain now preserves exhausted retry refusal through native API conversion and SQLite reload.", "regression_tests": [ "test_insights_v8_provider_attempt_deadline.py", "test_insights_shared_presentation_budget_review.py", "test_insights_pg_deadline.py" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json", "results-v8/v16-carry-pair-diagnosis/receipt.json", "results-v8/v16-carry-pair-diagnosis/repaired-review.json" ], "remaining_limits": "Remote resource use and multiprocess capacity are not proved. A provider exception alone does not prove a remote outage.", "final_result": "V16 frozen curated evidence; independent quality targets remain evaluator-owned" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V16 native six jobs finish; optional work does not block checked rows. Absolute provider HTTP deadlines, concurrent shared budgets and local cancellation have retained focused controls. The empty-Gold concurrent-provider drain now preserves exhausted retry refusal through native API conversion and SQLite reload.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Remote resource use and multiprocess capacity are not proved. A provider exception alone does not prove a remote outage.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F13", "title": "Entry-point agreement", "status": "partially_fixed", "problem": "Reused services did not ensure equal source, policy and result semantics.", "root_cause": "Some callers and serializers dropped actual source identity, unit_column or unit_unverified metadata. Shared execution services did not preserve those contracts by themselves.", "fix": "Propagate checked metadata through runtime summary callers, Compose, exports and shares.", "regression_tests": [ "test_full_result_export_enrichments.py", "test_bi_share_snapshot.py", "tests/js/bi-terminal-rendering.test.mjs" ], "references": [ "api/routers/bi_query.py:10929", "api/routers/bi_share.py:843" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Native Compose and refreshed CSV after source mutation remain unverified. Do not claim complete coverage of every required surface. Final limit: Partially fixed. Fresh native API, table, headline, snapshot CSV, authorized full-details share and reload agree. Compose and refreshed CSV after source changes remain unverified.", "final_result": "Partially fixed. Fresh native API, table, headline, snapshot CSV, authorized full-details share and reload agree. Compose and refreshed CSV after source changes remain unverified.", "current_evidence": "Fresh native API, table, headline, snapshot CSV, authorized full-details share and reload agree on checked values and scope. Two refusal outcomes agree across saved records and DOM.", "priority": "P2", "user_impact": "Reused services did not ensure equal source, policy and result semantics.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "tests/v7/affected.xml", "tests/v7/broader.xml", "review/v7-fixture-cleanup/v7-broader-classification.json", "evaluation/receipts/v7-final-check-summary.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "Original finding carried forward; relevant controls retained and challenged. Related R findings describe repairs made in this goal.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "partially_verified", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V5: 1,776 affected Python tests pass (57 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 skips (184 files). Source hashes match the frozen application.", "verification_snapshot": "final-lineage-v5", "references": [ "api/routers/bi_query.py:10831", "api/routers/bi_share.py:843" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json" ], "limits": "Intentional live refresh versus frozen snapshot behavior remains distinct." }, "historical_pre_v14_result_review": { "status": "partially_verified", "root_cause": "Some callers and serializers dropped actual source identity, unit_column or unit_unverified metadata. Shared execution services did not preserve those contracts by themselves.", "fix": "Propagate checked metadata through runtime summary callers, Compose, exports and shares.", "current_evidence": "The original controls remain in the final affected/broader and browser checks. Final V7: 2,234 affected Python tests pass (74 files; no skips). Broader: 3,704 pass, 10 original-revision failures, 13 live skips (184 files). Source hashes match frozen V7.", "regression_tests": [ "test_compose_bi_name_splice.py", "test_full_result_export_enrichments.py", "test_bi_share_snapshot.py" ], "limits": "Intentional live refresh versus frozen snapshot behavior remains distinct.", "final_result": "partially_verified" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_full_result_export_enrichments.py", "test_bi_share_snapshot.py", "tests/js/bi-terminal-rendering.test.mjs" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "V15 table, headline, CSV, authorized full-details share API/DOM and reload agree on 45 clients, GBP 2454.50 and count 45. Fresh native reads prove both refusal records agree. Exact byte identity reuses 244 Node and 72 mocked Chromium checks from V14.", "final_result": "V15_checked_facts_and_refusal_records_pass; transient_footer_wording_failed", "verification_snapshot": "final-contract-v7", "limits": "Full wording parity fails: the transient ready footer says checked rows for a zero-row refusal. Native cases do not certify refreshed CSV after data changes or native Compose.", "root_cause": "Some callers and serializers dropped actual source identity, unit_column or unit_unverified metadata. Shared execution services did not preserve those contracts by themselves.", "fix": "Propagate checked metadata through runtime summary callers, Compose, exports and shares." }, "v16_independent_result_disposition": { "id": "F13", "title": "Entry-point agreement", "status": "partially_fixed", "root_cause": "Some callers and serializers dropped actual source identity, unit_column or unit_unverified metadata. Shared execution services did not preserve those contracts by themselves.", "fix": "Propagate checked metadata through runtime summary callers, Compose, exports and shares.", "current_evidence": "V16 saved API, table, headline, snapshot CSV, authorized full-details share API/DOM and reload agree on45 clients, GBP2454.50 and count45. Both native refusals agree at top level and nested records. Neutral footer passes native long optional waits plus five mocked ready outcomes.", "regression_tests": [ "test_full_result_export_enrichments.py", "test_bi_share_snapshot.py", "tests/js/bi-terminal-rendering.test.mjs" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json", "results-v8/v17-rank-threshold-prepared/rank-threshold-red.json" ], "remaining_limits": "Native covers saved snapshot CSV and authorized full-details shares. Refreshed export after source changes and native Compose remain outside this gate. V16 development also proves an open display-cap regression: at least one settled booking serves only one of seven checked clients.", "final_result": "V16 native curated gate passed; later development wrong-success blocker remains open" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V16 saved API, table, headline, snapshot CSV, authorized full-details share API/DOM and reload agree on45 clients, GBP2454.50 and count45. Both native refusals agree at top level and nested records. Neutral footer passes native long optional waits plus five mocked ready outcomes.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Full wording parity fails: the transient ready footer says checked rows for a zero-row refusal. Native cases do not certify refreshed CSV after data changes or native Compose.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "F14", "title": "Independent answer evaluation", "status": "partially_fixed", "problem": "Passing structural tests did not establish correct values or model quality.", "root_cause": "Passing structural tests did not establish correct values or model quality. Remaining cause: The fixture sensitivity audit covered 38 registered controls but missed a ratified staff predicate. Numeric equality and text separation do not prove complete semantic fidelity.", "fix": "Add real PostgreSQL oracles, frozen corpus splits, counterexample fixtures, live model runs and paired scoring.", "regression_tests": [ "test_insights_goal2_eval.py", "test_insights_goal2_exports.py", "test_insights_postgres_values.py" ], "references": [ "eval_oracle/goal2.py:418" ], "evidence_type": "deterministic_regression", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "All three requested reviewers completed final challenges. Their reviews are bounded and are not an external audit. Full fixture predicate sensitivity is incomplete. All primary quality targets fail. The holdout will never be used for another candidate.", "final_result": "Partially fixed; independent quality certification is disproved. Counts and intervals reproduce. The fixture does not exercise every ratified predicate. All primary quality targets fail.", "current_evidence": "The repaired E5 baseline is complete. Final V18 is frozen and the sealed 250-attempt run is admitted. E6 is complete and failed. All final targets will retain fixed denominators. Final held evidence: Three final reviewers reproduce the metrics. Pipedrive employee values are 0 true, 52 false and 12 null; the ratified exclusion removes no rows.", "priority": "P2", "user_impact": "Passing structural tests did not establish correct values or model quality.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "evaluation/receipts/final-lineage-v5-primary-paired-v9.json", "evaluation/prospective-v10/orchestration/evidence/lead-admission.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json", "phase0/final-v18-late-predicate-fidelity-audit.json", "phase0/held-v18-lead-semantic-counterexamples.json" ], "resolution_provenance": "Original finding carried forward; relevant controls retained and challenged. Related R findings describe repairs made in this goal.", "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate.", "historical_v5": { "status": "partially_verified", "current_evidence": "The frozen V5 locked comparison failed the silent-wrong target: 37/245 before versus 39/245 after. Correct supported completions increased from 7/220 to 100/220. The paired error-reduction interval includes deterioration. A separate prospective 400-action corpus and 72-action supplement are sealed. No new provider admissions have started. Historical V5 suites do not verify current edits.", "verification_snapshot": "final-lineage-v5", "references": [ "eval_oracle/goal2.py:418" ], "evidence_links": [ "tests/final-affected.xml", "tests/final-broader.xml", "evaluation/receipts/v5-final-source-identity-and-deltas.json" ], "limits": "The new corpus has 137 unique question texts across 40 synthetic families. Its author saw V5 failures. It is not a pre-repair or author-blind holdout. Full response meaning, human usefulness, production mix and material improvement remain unverified." }, "historical_v7": { "status": "partially_verified", "final_result": "partially_verified", "current_evidence": "The complete V5 locked comparison failed the silent-wrong target: 37/245 before versus 39/245 after. Correct supported completions increased from 7/220 to 100/220. The separately sealed prospective 400-action comparison and 72-action supplement are now running on frozen V7. Outcomes stay unopened until the registered campaign ends. This row does not certify material improvement.", "verification_snapshot": "final-contract-v7", "current_source_verification": "Application bytes match the frozen V7 canonical map. All 2234 affected tests pass; broader 3704 pass, 10 original-revision failures and 13 configured live skips. Full mypy and changed-code Ruff pass. Test counts do not certify arbitrary model answers." }, "historical_pre_v14": { "limits": "The new corpus has 137 unique question texts across 40 synthetic families. Its author saw V5 failures. It is not a pre-repair or author-blind holdout. Full response meaning, human usefulness, production mix and material improvement remain unverified.", "status": "reopened", "final_result": "reopened" }, "status_before_current_final_verification": "reopened", "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "Repaired E5 completed250attempts;60/100firstturnerrors have product causes and user-authorized continuation. V15 complete70developmentpair/fullsuite/native are retained. E6 and independentcandidate250remain unadmitted.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "No quality claim cites the invalid oldv10run. The new holdout staysclosed; archive-policy annex changes no numeric values or scores.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." } }, { "id": "N01", "title": "Apply every ratified predicate", "status": "fixed", "problem": "Missing or unauthorized ratified columns can drop a definition predicate.", "root_cause": "Missing, restricted, renamed, and partially parameterized definition predicates are silently dropped by the compiler. The source parser also splits a three-part name incorrectly.", "fix": "Compile the exact ratified version against the authorized schema. Require all definition fields. Pass actual user and connection identities through each generation leg.", "regression_tests": [ "tests/test_insights_n01_definition_schema.py" ], "references": [], "evidence_type": "deterministic_diagnosis_and_retained_invalid_run", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "The proof covers the registered ratified definitions and finite schema controls. It does not prove every generated predicate. Final limit: Fixed within eight missing, restricted and renamed-column controls. A named definition cannot silently lose an unavailable column. Full fixture predicate sensitivity remains incomplete.", "final_result": "Fixed within eight missing, restricted and renamed-column controls. A named definition cannot silently lose an unavailable column. Full fixture predicate sensitivity remains incomplete.", "current_evidence": "Eight definition-schema controls pass for missing, restricted, renamed, and incomplete predicates. Both final arms use real definition retrieval and authorized compilation with the same deterministic adapter.", "priority": "P1", "user_impact": "Missing or unauthorized ratified columns can drop a definition predicate.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "query-v8/n01-before.xml", "query-v8/diagnostic-review.md", "query-v8/n01-after.xml", "phase0/development-r2-v9-paired/paired-summary.json", "results-v8/structural-browser-v9-results.json", "phase0/development-r2-v11-paired/paired-summary.json", "phase0/v11-cold-schema-credential-cause.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json", "phase0/final-v18-late-predicate-fidelity-audit.json", "phase0/held-v18-lead-semantic-counterexamples.json" ], "resolution_provenance": "New finding from the resumed user goal, registered before tuning.", "historical_pre_v13": { "status": "partially_fixed", "current_evidence": "V11 actual embedding search finds the certified private-client definition. Its cold inventory fails because product code requests the repository display password. The stored reader credential authenticates unchanged. V12 real encrypted-repository checks expose and repair this path plus sqlite3.Row cache-key normalization; all33 focused controls pass. Final native and live verification remain pending.", "final_result": "unverified_on_final_candidate", "verification_snapshot": "V11_f751e4cf_failed_development_and_V12_working_regressions" }, "historical_pre_v14": { "limits": "V12 is not frozen. Final native execution, independent evaluation, and isolated deterministic latency remain required.", "status": "partially_fixed", "final_result": "unverified_on_final_candidate" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V15 actual embedding retrieval and authorized definition compilation pass. Restricted, missing and renamed-column controls are bound by the final full-suite inventory. V16 remains under review.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Finite definition proof is structural. Final independent candidate run is unadmitted.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." }, "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate." }, { "id": "N02", "title": "Classify terminal failures", "status": "partially_fixed", "problem": "The generic failure message hides schema, provider, permission, and semantic causes.", "root_cause": "The legacy error model has no required class. Router catches and wrappers lose typed causes. Remaining cause: Several terminal paths retain a broad category without the precise generated-SQL cause. The original implementation still lacks runtime classes.", "fix": "A shared nine-class terminal contract now reaches API errors and persisted turns. A new append-only migration stores the class and recovery actions.", "regression_tests": [ "tests/test_insights_terminal_outcomes.py", "tests/test_insights_http_terminal_contract.py", "tests/test_insights_v9_contract_review.py", "tests/test_insights_v8_terminal_persistence.py" ], "references": [], "evidence_type": "deterministic_diagnosis_and_retained_invalid_run", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Original 21 runtime-unclassified error scores remain. A category-level message can still omit the detailed SQL or planning cause. The both-arm target cannot pass retroactively. Final limit: Partially fixed. Candidate terminal errors are classified and no outward generic crash text remains. Original has 21 unclassified errors; some candidate causes remain too general.", "final_result": "Partially fixed. Candidate terminal errors are classified and no outward generic crash text remains. Original has 21 unclassified errors; some candidate causes remain too general.", "current_evidence": "Final terminal contract controls pass. Native write and grain refusals preserve class and outcome. The legacy generic outward error sentence is replaced by a symmetric presentation adapter in both evaluation arms. Final held evidence: The result reviewer checks 117 answer records and 114 present confidence records: zero outcome parity mismatches, all 117 classified, zero generic crashes.", "priority": "P1", "user_impact": "The generic failure message hides schema, provider, permission, and semantic causes.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "n02-n03-diagnostic-red.xml", "phase0/projection-audit-v2.xml", "query-v8/generation-terminal-after.xml", "phase0/development-r2-v9-paired/paired-summary.json", "results-v8/structural-browser-v9-results.json", "query-v8/interim-v9-contract-red.xml", "query-v8/interim-v10-contract-expanded.xml", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "New finding from the resumed user goal, registered before tuning.", "historical_pre_v13": { "status": "partially_fixed", "current_evidence": "The frozen V10 browser exposed raw SQL validation text as the CT user cause. Independent review also found quoted credential strings and full-export terminal classes lost under HTTP502. V11 centralizes the CT cause and preserves the typed export envelope. The expanded root controls pass; final source-wide and browser verification remain pending.", "final_result": "unverified_on_final_candidate", "verification_snapshot": "V9_45497f40_development_and_V10_working_regressions" }, "historical_pre_v14": { "limits": "V10 is not frozen. Native end-to-end checks, final independent evaluation, and final deterministic p95 remain required.", "status": "partially_fixed", "final_result": "unverified_on_final_candidate" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V15 has zero unclassified candidate task errors across70development attempts. Five task errors retain native classes. Original21unclassified scores remain unchanged. One sql-labelled merge failure has a separate semantic-cause audit.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Original errors cannot be retroactively changed to satisfy the both-arm target.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." }, "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate." }, { "id": "N03", "title": "Render safe refusals as answers", "status": "partially_fixed", "problem": "Grain, unit, calendar, and write refusals can appear as execution errors.", "root_cause": "A correct guard refusal is stored or rendered with the legacy execution-error path.", "fix": "Typed guard outcomes now retain refusal status and suggested rephrases. The browser uses the refusal outcome even on old error-status records.", "regression_tests": [ "tests/test_insights_terminal_outcomes.py", "tests/test_insights_v18_money_guard_recovery.py" ], "references": [], "evidence_type": "deterministic_diagnosis_and_retained_invalid_run", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "A safe refusal does not earn completion credit for a supported question. Live unresolved refusals remain in the final denominators. Final limit: Partially fixed. Native permission and grain refusals carry a class, reason and suggested question. All 117 candidate terminal records have matching outcome projections. The 24 held refusals do not earn completion credit.", "final_result": "Partially fixed. Native permission and grain refusals carry a class, reason and suggested question. All 117 candidate terminal records have matching outcome projections. The 24 held refusals do not earn completion credit.", "current_evidence": "V18 moves money guards into the existing bounded semantic correction path and preserves typed refusals after exhaustion. Native write and grain refusals agree across saved and rendered records.", "priority": "P1", "user_impact": "Grain, unit, calendar, and write refusals can appear as execution errors.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "query-v8/projection-audit.md", "phase0/projection-audit-final.xml", "results-v8/terminal-outcomes-light.png", "results-v8/terminal-outcomes-dark.png", "phase0/development-r2-v9-paired/paired-summary.json", "results-v8/structural-browser-v9-results.json", "query-v8/interim-v9-contract-red.xml", "query-v8/interim-v10-contract-expanded.xml", "phase0/development-r2-v11-paired/paired-summary.json", "phase0/v11-cold-schema-credential-cause.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "New finding from the resumed user goal, registered before tuning.", "historical_pre_v13": { "status": "partially_fixed", "current_evidence": "V10 native write refusal passes; no native grain pass is claimed. V11 explicit write/secret requests refuse before dispatch, but the early conversation path loses the actual guard/code from its saved response. Frozen scoring keeps refusal_unverified. V12 adds native durable provenance from the actual policy decision; no scorer or historical label changes.", "final_result": "unverified_on_final_candidate", "verification_snapshot": "V11_f751e4cf_failed_development_and_V12_working_regressions" }, "v13_native_evidence": "Separate duplicate-trade fixture produces explicit grain refusal with real guard, code, reason and rephrase. No error projection is installed.", "historical_pre_v14": { "limits": "V12 is not frozen. Final native execution, independent evaluation, and isolated deterministic latency remain required.", "status": "partially_fixed", "final_result": "unverified_on_final_candidate" }, "history_before_v18_boundary_diagnosis": { "current_evidence": "V15 all20refusals agree across saved top-level and both answer records. Native write and grain refusals pass fresh API/DOM parity. Transient checked-rows wording still fails; V16 neutral footer checks are in progress." }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V15 all20refusals agree across saved top-level and both answer records. Native write and grain refusals pass fresh API/DOM parity. Transient checked-rows wording still fails; V16 neutral footer checks are in progress. V17 live development: 20/70 correct, 7/28 first-turn correct, 1/28 first-turn error, 0/14 full conversations. Final independent evaluation remains closed. Reopened: early native pipeline currency guard used a generic validation error on b2; the share inherited it on b4. V18 moves money guards into the existing bounded semantic correction path. Verification is in progress.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Safe guard refusal is distinct from correct completion of a supported question. All18supportedrefusals remain in the score.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." }, "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate." }, { "id": "N04", "title": "Execute supported first turns", "status": "partially_fixed", "problem": "Supported first-turn questions fail in the repaired original application.", "root_cause": "60/100 supported first turns failed:36 unusable leg plans,10 unavailable generated schema references,7 missing merge keys,5 missing source qualification,1 empty generation,1 wrong function input type. All60 are product causes. Remaining cause: Generated SQL, source selection and result-shape planning remain unreliable on unfamiliar first-turn questions.", "fix": "Product fixes target unusable plan SQL, fabricated schema names, missing join keys, source qualification, and wrong function input types.", "regression_tests": [ "tests/test_insights_cold_schema_context.py", "tests/test_insights_v18_money_guard_recovery.py" ], "references": [], "evidence_type": "live_original_baseline_interim_gate", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "V15 development first-turn correct was 10/28; V17 was 7/28. Both were below 40%. Final held metrics remain separate. Final limit: Partially fixed. First-turn execution errors fall from 60/100 to 35/100. Correct first turns rise from 14/100 to 18/100. Both absolute targets fail.", "final_result": "Partially fixed. First-turn execution errors fall from 60/100 to 35/100. Correct first turns rise from 14/100 to 18/100. Both absolute targets fail.", "current_evidence": "E5 has 60 product-caused first-turn errors out of 100. Final source guards and execution controls pass, but the held run still exposes source-routing and generated SQL failures. Final held evidence: Completed E5 cause audit attributes all 60 original first-turn errors to product causes. V18 first-turn errors are 18 SQL, nine schema and eight semantic.", "priority": "P1", "user_impact": "Supported first-turn questions still fail or refuse. The repaired environment exposes product failures.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "phase0/fixture-self-check.json", "phase0/definition-retrieval-original.json", "phase0/e5-start.json", "phase0/e5-threshold-crossed.json", "phase0/e5-summary-sanitized.json", "phase0/e5-case-receipts-anonymous.json", "phase0/development-r2-v9-paired/paired-summary.json", "results-v8/structural-browser-v9-results.json", "phase0/development-r2-v11-paired/paired-summary.json", "phase0/v11-cold-schema-credential-cause.json", "tests-v17/scoped/receipt.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "New finding from the resumed user goal, registered before tuning.", "historical_pre_v13": { "status": "partially_fixed", "current_evidence": "Complete V11 development:28/28 first-turn errors,0/28 first-turn correct;70 attempts retained,zero provider calls. All failures before generation trace to product masked-password loading in the new cold inventory. The actual reader credential works. V12 fixes that seam and real-row validator cache identity. This result does not invalidate fixture or credential setup.", "final_result": "unverified_on_final_candidate", "verification_snapshot": "V11_f751e4cf_failed_development_and_V12_working_regressions" }, "historical_pre_v14": { "limits": "V12 is not frozen. Final native execution, independent evaluation, and isolated deterministic latency remain required.", "status": "partially_fixed", "final_result": "unverified_on_final_candidate" }, "historical_before_v17_lead_update": { "current_evidence": "V15 first-turn errors2/28 and correct10/28. Both recurring first-turn errors lack usable Gold SQL for a live booking condition. V16 tests the existing intersection source contract and preserves the bound-date execution route.", "final_result": "V15 evidence retained; V16 review and final independent acceptance pending.", "verification_snapshot": "V12 measured; V13 working scoped checks, final unverified", "limits": "Opened development does not certify independent first-turn accuracy.", "root_cause": "60/100 supported first turns failed:36 unusable leg plans,10 unavailable generated schema references,7 missing merge keys,5 missing source qualification,1 empty generation,1 wrong function input type. All60 are product causes.", "fix": "Product fixes target unusable plan SQL, fabricated schema names, missing join keys, source qualification, and wrong function input types." }, "history_before_v18_boundary_diagnosis": { "current_evidence": "Original E5 has 60 product-caused first-turn errors out of 100; zero fixture or environment causes. User accepted the environment and authorized tuning. V16 opened development has 1/28 first-turn errors but only 6/28 first-turn correct. Final holdout unadmitted. V17 final scoped gate passes 701 checks with no skips; native bounded correction returns both retained output cases correctly. Frozen full/native/live verification is pending." }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "Original E5 has 60 product-caused first-turn errors out of 100; zero fixture or environment causes. User accepted the environment and authorized tuning. V16 opened development has 1/28 first-turn errors but only 6/28 first-turn correct. Final holdout unadmitted. V17 final scoped gate passes 701 checks with no skips; native bounded correction returns both retained output cases correctly. Frozen full/native/live verification is pending. V17 live development: 20/70 correct, 7/28 first-turn correct, 1/28 first-turn error, 0/14 full conversations. Final independent evaluation remains closed.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Opened development does not certify independent first-turn accuracy.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." }, "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate." }, { "id": "N05", "title": "Prevent correct-answer regressions", "status": "partially_fixed", "problem": "Three retained v10 original-correct cases became task errors.", "root_cause": "Write intent after a scope clause escapes the refusal gate. Scalar gold-leg planner support is not wired in the router. Other per-case causes remain under diagnosis. Remaining cause: New routing, shape and guard decisions can reject or misinterpret questions the original answered correctly.", "fix": "Scope-prefixed write checks and Gold final-answer routing are implemented. The paired regression gate retains every original outcome and unavailable export action.", "regression_tests": [ "tests/test_insights_v8_intent_shape_diagnostics.py", "test_insights_v17_rank_thresholds.py" ], "references": [], "evidence_type": "deterministic_diagnosis_and_retained_invalid_run", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "All losses to refusal, blocking and clarification must be shown beside the narrow correct-to-error-or-wrong gate. Final limit: Partially fixed; the paired regression gate fails. Seven original correct answers become three errors and four wrong answers. Only 24/36 original completions are retained; four refusals and one clarification account for the other five losses.", "final_result": "Partially fixed; the paired regression gate fails. Seven original correct answers become three errors and four wrong answers. Only 24/36 original completions are retained; four refusals and one clarification account for the other five losses.", "current_evidence": "The three retained v10 regression families have scoped repairs. V17 preserved the sole correct original development case. Final V18 paired transition and completion-retention gates are required. Final held evidence: The complete retention table lists all 12 lost original completions. No refusal, clarification or blocked turn is counted as a retained answer.", "priority": "P1", "user_impact": "Three retained v10 original-correct cases became task errors.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "query-v8/v10-selected-counterexamples.json", "results-v8/diagnostic-review.md", "phase0/development-r2-v9-paired/paired-summary.json", "results-v8/structural-browser-v9-results.json", "tests-v17/scoped/receipt.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "New finding from the resumed user goal, registered before tuning.", "historical_pre_v13": { "status": "partially_fixed", "current_evidence": "Repaired original versus V10 development has zero narrow correct-to-error-or-wrong transitions. The original lone correct case becomes blocked_by_parent. It is no longer completed. None of the32 selected historical regression actions is independently verified fixed by this run.", "final_result": "unverified_on_final_candidate", "verification_snapshot": "V9_45497f40_development_and_V10_working_regressions" }, "historical_pre_v14": { "limits": "V10 is not frozen. Native end-to-end checks, final independent evaluation, and final deterministic p95 remain required.", "status": "partially_fixed", "final_result": "unverified_on_final_candidate" }, "historical_before_v17_lead_update": { "current_evidence": "V15 retains the sole original correct development answer with zero correct-to-error/wrong transitions. The all70pair and selected32gate remain. Final250case regression gate is unrun.", "final_result": "V15 evidence retained; V16 review and final independent acceptance pending.", "verification_snapshot": "V12 measured; V13 working scoped checks, final unverified", "limits": "Finite numerical equality is audited for semantic counterexamples; it is not full meaning proof.", "root_cause": "Write intent after a scope clause escapes the refusal gate. Scalar gold-leg planner support is not wired in the router. Other per-case causes remain under diagnosis.", "fix": "Scope-prefixed write checks and Gold final-answer routing are implemented. The paired regression gate retains every original outcome and unavailable export action." }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V16 preserves the sole correct original development case. The retained inactive membership case still needs V17 full/native/live verification after the threshold repair. Final 250-case paired gate unadmitted. V17 final scoped gate passes 701 checks with no skips; native bounded correction returns both retained output cases correctly. Frozen full/native/live verification is pending.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Finite numerical equality is audited for semantic counterexamples; it is not full meaning proof.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." }, "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate." }, { "id": "N06", "title": "Repair newly executable wrong answers", "status": "partially_fixed", "problem": "Twenty-four v10 original-error cases became silent wrong.", "root_cause": "Source membership, union and intersection routes, missing values, and tie semantics share query-meaning failures. Remaining cause: Semantic guards do not fully bind explicit question predicates, temporal operators and output operations to the executed SQL.", "fix": "Counterexamples are retained. V10 implements proved identity bridging and source membership fixes, with exact-value PostgreSQL controls in progress.", "regression_tests": [ "test_insights_v17_membership_predicates.py", "tests/test_insights_v18_membership_output.py", "tests/test_insights_v8_ct_input_regressions.py" ], "references": [], "evidence_type": "deterministic_diagnosis_and_retained_invalid_run", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Two numerically correct held first turns also have proved semantic counterexamples. No score or application change follows this diagnosis. Final limit: Partially fixed. Silent wrong answers rise from 31/250 to 54/250. Wrong filters, earliest-date logic, source membership and output shape remain.", "final_result": "Partially fixed. Silent wrong answers rise from 31/250 to 54/250. Wrong filters, earliest-date logic, source membership and output shape remain.", "current_evidence": "The retained 24 new-wrong cases have counterexamples and shared source/membership repairs. V18 fixes NULL-name predicate bypass and requested membership projection. The held run still contains wrong source and output shapes. Final held evidence: Independent review reproduces dropped marketing eligibility, earliest-to-any verification date and scalar counts returned as per-client listings.", "priority": "P1", "user_impact": "Twenty-four v10 original-error cases became silent wrong.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "query-v8/diagnostic-review.md", "phase0/development-r2-v9-paired/paired-summary.json", "results-v8/structural-browser-v9-results.json", "tests-v17/scoped/receipt.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "New finding from the resumed user goal, registered before tuning.", "historical_pre_v13": { "status": "partially_fixed", "current_evidence": "V10 completes no scored correct or wrong answers in70 attempts. Its apparent100% wrong-answer reduction is caused by failure, refusal and blocking; it is not useful correctness. Shared identity/membership repairs have focused PostgreSQL controls, but final live repair evidence is open.", "final_result": "unverified_on_final_candidate", "verification_snapshot": "V9_45497f40_development_and_V10_working_regressions" }, "historical_pre_v14": { "limits": "V10 is not frozen. Native end-to-end checks, final independent evaluation, and final deterministic p95 remain required.", "status": "partially_fixed", "final_result": "unverified_on_final_candidate" }, "historical_before_v17_lead_update": { "current_evidence": "V15 two wrong successes: written top-five limit becomes38rows; professional-referral membership gains extra conditions and returns0instead of11. One correct empty result uses the wrong manual-origin literal; two unverified comparisons retain physical-source defects. V16 fixes are under review.", "final_result": "V15 evidence retained; V16 review and final independent acceptance pending.", "verification_snapshot": "V12 measured; V13 working scoped checks, final unverified", "limits": "All old scores and counterexamples remain. Domain-policy discrepancies have explicit separate reconciliation.", "root_cause": "Source membership, union and intersection routes, missing values, and tie semantics share query-meaning failures.", "fix": "Counterexamples are retained. V10 implements proved identity bridging and source membership fixes, with exact-value PostgreSQL controls in progress." }, "history_before_v18_boundary_diagnosis": { "current_evidence": "V17 removes a proved unrequested snapshot-affiliation predicate and applies the settled-membership deletion default. Native counterexamples preserve stale-snapshot client 1, exclude deleted-only client 3, and keep realized SUM 210. The boundary gate passes 47 checks. V17 final scoped gate passes 701 checks with no skips; native bounded correction returns both retained output cases correctly. Frozen full/native/live verification is pending." }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V17 removes a proved unrequested snapshot-affiliation predicate and applies the settled-membership deletion default. Native counterexamples preserve stale-snapshot client 1, exclude deleted-only client 3, and keep realized SUM 210. The boundary gate passes 47 checks. V17 final scoped gate passes 701 checks with no skips; native bounded correction returns both retained output cases correctly. Frozen full/native/live verification is pending. V17 live development: 20/70 correct, 7/28 first-turn correct, 1/28 first-turn error, 0/14 full conversations. Final independent evaluation remains closed. V18 native intersection reproduces V17 final alias refusal and verifies requested client_id/name, display ordering before capping, and complete checked carry. Missing name remains a refusal.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "All old scores and counterexamples remain. Domain-policy discrepancies have explicit separate reconciliation.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." }, "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate." }, { "id": "N07", "title": "Bound attempts and provider retries", "status": "partially_fixed", "problem": "Five V7 jobs reached the 180-second deadline.", "root_cause": "Each of the five timeout jobs spent about 179 seconds in its first failed provider call. The request lacks an absolute deadline within the job budget. Remaining cause: Bounded attempts stop parent deadlines but do not remove slow generation or optional cancellation. SDK errors do not by themselves establish a remote outage.", "fix": "All provider calls share a six-attempt turn budget and one provider retry. The HTTP transport now enforces an absolute request deadline. Propagate the typed CT future failure when Gold returns no usable SQL.", "regression_tests": [ "test_insights_v8_provider_attempt_deadline.py", "test_insights_shared_presentation_budget_review.py", "test_insights_v16_provider_drain.py" ], "references": [], "evidence_type": "deterministic_diagnosis_and_retained_invalid_run", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "A failed SDK call may reflect provider availability, a timeout, or cancellation. Remote cause is not proved by exception type alone. E6 p95 fails. Final limit: Partially fixed. All 250 candidate jobs finish without a deadline. SQL generation makes 453 boundary calls, with 11 SDK failures; 118 auxiliary calls have nine local cancellation failures. E6 p95 fails.", "final_result": "Partially fixed. All 250 candidate jobs finish without a deadline. SQL generation makes 453 boundary calls, with 11 SDK failures; 118 auxiliary calls have nine local cancellation failures. E6 p95 fails.", "current_evidence": "The final controls bound total generation and one provider retry per turn. E6 confirms all deterministic external calls. Native jobs finish and the held provider/job ledger is retained. Final held evidence: The independent reviewer recomputes all 180 E6 timings. Candidate p95 is 1732.008 ms versus 1549.449 ms. No failed or refused timing is removed.", "priority": "P1", "user_impact": "Five V7 jobs reached the 180-second deadline.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "results-v8/diagnostic-review.md", "query-v8/generation-terminal-after.xml", "phase0/development-r2-v9-paired/paired-summary.json", "results-v8/structural-browser-v9-results.json", "phase0/development-r2-v11-paired/paired-summary.json", "phase0/v11-cold-schema-credential-cause.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "New finding from the resumed user goal, registered before tuning.", "historical_pre_v13": { "status": "partially_fixed", "current_evidence": "V11 has zero external generation calls and no deadline jobs; it cannot prove useful workload latency. The final V11 review reproduced six cancellation wraps in three CT dispatch paths and passed all53 repair/neighbor checks after the fix. Shared optional work retains the parent six-attempt and one-retry budget. Final isolated deterministic latency remains required.", "final_result": "unverified_on_final_candidate", "verification_snapshot": "V11_f751e4cf_failed_development_and_V12_working_regressions" }, "historical_pre_v14": { "limits": "V12 is not frozen. Final native execution, independent evaluation, and isolated deterministic latency remain required.", "status": "partially_fixed", "final_result": "unverified_on_final_candidate" }, "historical_pre_v14_result_review": { "status": "partially_fixed", "root_cause": "Each of the five timeout jobs spent about 179 seconds in its first failed provider call. The request lacks an absolute deadline within the job budget.", "fix": "All provider calls share a six-attempt turn budget and one provider retry. The HTTP transport now enforces an absolute request deadline.", "current_evidence": "V12 has zero180-second deadlines but31provider timeouts and3connection failures in101SQL calls. Fixed six-call serial diagnosis reproduces one25-second provider timeout; five finish. Source of remote delay remains unresolved. Two optional25-second cancellations are separately classified and retained.", "regression_tests": [ "tests/test_insights_v8_provider_attempt_deadline.py" ], "limits": "V14 is under review. Its final source freeze, native checks, independent paired evaluation, and isolated deterministic latency are pending. Earlier candidate results remain separate.", "final_result": "unverified_on_final_candidate" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json", "results-v8/v14-auxiliary-deadline-independent-review.json" ], "v14_regression_tests": [ "test_insights_v8_provider_attempt_deadline.py", "test_insights_shared_presentation_budget_review.py" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v14-auxiliary-deadline-independent-review.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "V16 fixes the empty-Gold branch that discarded the concurrent CT provider failure. Eight valid controls fail before and pass after. The 67-check provider/neighbor gate passes without skips. Independent actual SQL generator, API and SQLite checks retain one shared retry and a saved provider refusal after exactly two failed calls. Final live deadline and provider-availability rates remain unmeasured.", "final_result": "unverified_on_final_candidate", "verification_snapshot": "V16 scoped source; final frozen run pending", "limits": "Provider availability is not configuration validity. The isolated fixed-provider latency run is still unadmitted.", "root_cause": "Each of the five timeout jobs spent about 179 seconds in its first failed provider call. The request lacks an absolute deadline within the job budget.", "fix": "All provider calls share a six-attempt turn budget and one provider retry. The HTTP transport now enforces an absolute request deadline. Propagate the typed CT future failure when Gold returns no usable SQL." }, "v16_independent_result_disposition": { "id": "N07", "title": "Bound attempts and provider retries", "status": "fixed_on_curated_controls", "root_cause": "Each of the five timeout jobs spent about 179 seconds in its first failed provider call. The request lacks an absolute deadline within the job budget.", "fix": "All provider calls share a six-attempt turn budget and one provider retry. The HTTP transport now enforces an absolute request deadline.", "current_evidence": "Shared generation attempts and one provider retry include optional children. V16 native six jobs finish with no deadline. Independent actual-generator -> planner -> API -> SQLite controls preserve exhausted provider retry as refusal, classprovider, codeBI_PROVIDER_UNAVAILABLE. Historical V15 auxiliary25-second failure is retained separately. V16 live and independent E6 results are root/evaluator-owned and not claimed here.", "regression_tests": [ "test_insights_v8_provider_attempt_deadline.py", "test_insights_shared_presentation_budget_review.py" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v14-auxiliary-deadline-independent-review.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json", "results-v8/v16-carry-pair-diagnosis/receipt.json", "results-v8/v16-carry-pair-diagnosis/repaired-review.json", "results-v8/v15-optional-failure-attribution.json" ], "remaining_limits": "Live inventories count complete/execute boundaries, not every SDK HTTP retry. Remote termination is not proved. Historical V15 selected-membership PlannerError cannot be assigned to the separately reproduced empty-Gold drain. E6 and final independent deadline rates remain separate.", "final_result": "V16 frozen curated evidence; independent quality targets remain evaluator-owned" }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "Shared generation attempts and one provider retry include optional children. V16 native six jobs finish with no deadline. Independent actual-generator -> planner -> API -> SQLite controls preserve exhausted provider retry as refusal, classprovider, codeBI_PROVIDER_UNAVAILABLE. Historical V15 auxiliary25-second failure is retained separately. V16 live and independent E6 results are root/evaluator-owned and not claimed here.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Provider availability is not configuration validity. The isolated fixed-provider latency run is still unadmitted.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." }, "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate." }, { "id": "N08", "title": "Preserve and recover conversation scope", "status": "partially_fixed", "problem": "Both v10 arms completed zero of 80 conversations.", "root_cause": "The anchor search skips a blocked selected parent and can reuse an older group. Sibling how-many follow-ups also conflict with ratified listing continuity. Remaining cause: The reference recognizers miss explicit variants such as first checked result and checked IDs. These turns bypass blocked-parent recovery and checked membership carry.", "fix": "A non-successful selected parent blocks older-group carry. Recovery offers a rerun or a named group. Successful saved counts retain exact checked membership. Mask only the proved aggregate NULL-policy group token before source recognition, preserving offsets and genuine account-group terms.", "regression_tests": [ "test_bi_followup_router.py", "test_insights_v16_aggregate_group_scope.py", "test_insights_v17_output_recovery_native.py", "test_insights_v8_chain_recovery.py", "test_insights_v8_context_contract.py", "tests/test_insights_v18_membership_output.py" ], "references": [], "evidence_type": "deterministic_diagnosis_and_retained_invalid_run", "confidence": "Source-bound tests and bounded native observations. All three final independent reviews are complete. Held quality and fixture sensitivity have material limits.", "limits": "Full held conversation success is required. Wrong parent answers can propagate a checked but semantically wrong group. Final limit: Partially fixed. Fifty-seven curated chain controls pass, but held full-conversation success is only 4/50 versus 1/50. All 44 explicit continuations after failed/refused parents bypass recovery: 31 errors, nine refusals and four wrong answers.", "final_result": "Partially fixed. Fifty-seven curated chain controls pass, but held full-conversation success is only 4/50 versus 1/50. All 44 explicit continuations after failed/refused parents bypass recovery: 31 errors, nine refusals and four wrong answers.", "current_evidence": "All 57 three/five-turn recovery controls pass, including a failure at each position. The fresh native three-turn chain carries exactly 45 clients. Final held evidence: The result reviewer calls both frozen scope detectors for all 44 child questions. Both detectors return false for every checked first/third-result or ID reference.", "priority": "P1", "user_impact": "Both v10 arms completed zero of 80 conversations.", "verification_snapshot": "Final V18 8958ae93. Full suite, native browser, isolated E6 and the sole 250-attempt paired holdout are complete. No post-holdout tuning.", "evidence_links": [ "results-v8/diagnostic-review.md", "phase0/development-r2-v9-paired/paired-summary.json", "results-v8/structural-browser-v9-results.json", "query-v8/interim-v9-contract-red.xml", "query-v8/interim-v10-contract-expanded.xml", "phase0/development-r2-v11-paired/paired-summary.json", "phase0/v11-cold-schema-credential-cause.json", "tests-v17/scoped/receipt.json", "tests-v18/final-curated-and-matrix-binding.json", "results-v8/v18-native-acceptance-input.json", "phase0/e6-final-v18/run/summary.json", "phase0/final-v18-heldout/paired-summary.json", "phase0/final-v18-acceptance.json", "phase0/final-v18-evaluator-review.json", "query-v8/final-v18-query-review.json", "results-v8/final-v18-result-review.json" ], "resolution_provenance": "New finding from the resumed user goal, registered before tuning.", "historical_pre_v13": { "status": "partially_fixed", "current_evidence": "V11 development has0/14 complete conversations and26 blocked-by-parent turns after its credential-loading failures. Native first listing fails before source SQL dispatch. V12 also fixes three complete fresh-question phrases that the UI appended to an unanswered clarification;16 actual Chromium submission checks pass. Final correct-parent carry still needs native and live proof.", "final_result": "unverified_on_final_candidate", "verification_snapshot": "V11_f751e4cf_failed_development_and_V12_working_regressions" }, "v13_native_evidence": "Correct native45-client listing,2454.50total and45count;30final reads dispatch no execution. Blocked parent offers minimal recovery.", "historical_pre_v14": { "limits": "V12 is not frozen. Final native execution, independent evaluation, and isolated deterministic latency remain required.", "status": "partially_fixed", "final_result": "unverified_on_final_candidate" }, "historical_pre_v14_result_review": { "status": "partially_fixed", "root_cause": "The anchor search skips a blocked selected parent and can reuse an older group. Sibling how-many follow-ups also conflict with ratified listing continuity.", "fix": "A non-successful selected parent blocks older-group carry. Recovery offers a rerun or a named group. Successful saved counts retain exact checked membership.", "current_evidence": "Native V12 three-turn successful chain preserves45checkedmembers. Opened development remains0/14 full-conversation success and1/13 unnecessary clarification after successful parents. V13 fixes ranked-book scope release; blocked-parent and longer-chain final checks remain required.", "regression_tests": [ "tests/test_insights_v8_chain_recovery.py", "tests/test_insights_v9_contract_review.py", "tests/test_insights_v10_member_lineage.py" ], "limits": "V14 is under review. Its final source freeze, native checks, independent paired evaluation, and isolated deterministic latency are pending. Earlier candidate results remain separate.", "final_result": "unverified_on_final_candidate" }, "v14_review_evidence": [ "results-v8/structural-browser-v14.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v14-integrated-independent-review-frozen.json" ], "v14_regression_tests": [ "test_bi_followup_router.py", "test_insights_v8_chain_recovery.py", "test_insights_v8_context_contract.py" ], "current_receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json" ], "historical_before_v17_lead_update": { "current_evidence": "V16 native follow-up retains all seven checked keys and returns exact three currency sums using Gold only. The aggregate NULL-policy word group no longer creates a CT account-group leg. Fifty carry/routing/native checks pass with no skips. Final live conversation success remains unmeasured.", "final_result": "unverified_on_final_candidate", "verification_snapshot": "V16 scoped source; final frozen run pending", "limits": "Curated carry passes do not establish full live conversation success.", "root_cause": "The anchor search skips a blocked selected parent and can reuse an older group. Sibling how-many follow-ups also conflict with ratified listing continuity.", "fix": "A non-successful selected parent blocks older-group carry. Recovery offers a rerun or a named group. Successful saved counts retain exact checked membership. Mask only the proved aggregate NULL-policy group token before source recognition, preserving offsets and genuine account-group terms." }, "v16_independent_result_disposition": { "id": "N08", "title": "Preserve and recover conversation scope", "status": "partially_fixed", "root_cause": "The anchor search skips a blocked selected parent and can reuse an older group. Sibling how-many follow-ups also conflict with ratified listing continuity.", "fix": "A non-successful selected parent blocks older-group carry. Recovery offers a rerun or a named group. Successful saved counts retain exact checked membership.", "current_evidence": "V16 native listing -> total -> count retains all45 keys; refused write -> follow-up offers rerun or named group. Separate deterministic three/five-turn chains cover a failure at each position. Retained V15 seven-key pipeline continuation proved an aggregate NULL-policy group was misread as an account dimension; V16 masks only that proved phrase before recognition and preserves real/quoted group meanings. This routing repair has actual Gold final aggregation over seven keys; general conversation success awaits evaluation.", "regression_tests": [ "test_bi_followup_router.py", "test_insights_v8_chain_recovery.py", "test_insights_v8_context_contract.py" ], "receipts": [ "results-v8/structural-browser-v15.md", "results-v8/v14-result-repair-green.txt", "results-v8/v13-auxiliary-deadline-independent-review.json", "results-v8/ui-verification-v14/receipt.json", "results-v8/v15-integrated-independent-review-frozen.json", "results-v8/v15-native-acceptance-input.json", "results-v8/v15-ui-verification-byte-identity.json", "results-v8/structural-browser-v16.md", "results-v8/v16-native-acceptance-input.json", "results-v8/v16-ui-verification-source-binding.json", "results-v8/v16-carry-pair-diagnosis/receipt.json", "results-v8/v16-carry-pair-diagnosis/repaired-review.json", "results-v8/v17-rank-threshold-prepared/rank-threshold-red.json" ], "remaining_limits": "Curated chains do not establish independent full-conversation success. Exact seven-key grammar is backend/persistence proof, not an added native browser case. Fresh-question and offered-choice submission controls pass in current mocked Chromium. Written ranking counts are bounded to99. Compound top/bottom slice correctness is not claimed. Historical membership PlannerError cause remains unresolved; the aggregate-group routing defect is independently proved. V16 development also proves an open display-cap regression: at least one settled booking serves only one of seven checked clients.", "final_result": "V16 native curated gate passed; later development wrong-success blocker remains open" }, "history_before_v18_boundary_diagnosis": { "current_evidence": "V16 native three-turn chain carries 45 clients, GBP2454.50 and count45. Opened development remains 0/14 full conversations with 16 blocked follow-ups; unnecessary clarification is 0/10 after successful parents. V17 threshold/output repairs need fresh live carry proof. V17 final scoped gate passes 701 checks with no skips; native bounded correction returns both retained output cases correctly. Frozen full/native/live verification is pending." }, "history_before_final_v18": { "status": "partially_fixed", "current_evidence": "V16 native three-turn chain carries 45 clients, GBP2454.50 and count45. Opened development remains 0/14 full conversations with 16 blocked follow-ups; unnecessary clarification is 0/10 after successful parents. V17 threshold/output repairs need fresh live carry proof. V17 final scoped gate passes 701 checks with no skips; native bounded correction returns both retained output cases correctly. Frozen full/native/live verification is pending. V17 live development: 20/70 correct, 7/28 first-turn correct, 1/28 first-turn error, 0/14 full conversations. Final independent evaluation remains closed. V18 native intersection reproduces V17 final alias refusal and verifies requested client_id/name, display ordering before capping, and complete checked carry. Missing name remains a refusal.", "final_result": "Final integrated candidate and independent acceptance remain unverified. V17 results are retained development evidence.", "limits": "Curated carry passes do not establish full live conversation success.", "verification_snapshot": "V17 full, native and opened development completed. Working V18 corrections are not frozen." }, "current_source_verification": "Final source manifest 39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165 covers 1,692 application files. Historical source line bindings remain labeled by their original candidate." } ], "status_legend": { "fixed_in_curated_checks": "Fixed for the reproduced cases; no arbitrary-query guarantee.", "partially_verified": "Partially fixed or verified; the stated limit remains.", "disproved": "The current code did not reproduce the claimed failure.", "unverified": "The required evidence is unavailable.", "partially_fixed": "Stated counterexamples have bounded evidence; a current verification or outcome gap remains." }, "source_scope": "Final application source is bound by the V18 manifest. Earlier reference/source line fields are historical V7 evidence only. Final test binding is tests-v18/final-curated-and-matrix-binding.json.", "active_checkpoint": "e3516936afff335630241ac98726305dce4c1261", "retained_v7_matrix": "finding-matrix-v7-retained.json", "application_checkpoint": "8958ae93ff196bdc5fbcfd08ae12f81445c5d28c", "application_manifest_scope": "Exact frozen V17 source. V18 boundary corrections are working changes and need integrated verification.", "evaluation_amendment": "docs/plans/2026-09-06-insights-valid-evaluation-amendment.md", "active_candidate_checkpoint": "8958ae93ff196bdc5fbcfd08ae12f81445c5d28c", "v12_measurements": { "scope": "opened development only; independent candidate holdout closed", "attempts": 70, "supported_first_turns": 28, "task_error": 9, "silent_wrong": 3, "correct_completed": 4, "first_turn_errors": 4, "first_turn_correct": 4, "precision_numerator": 4, "precision_denominator": 7, "full_conversation_success": 0, "conversations": 14, "deadline_jobs": 0, "provider_calls": 101, "provider_timeouts": 31, "provider_connection_failures": 3, "environment_configuration": "valid; provider availability incidents remain an explicit failed acceptance target", "evidence": "phase0/development-r2-v12-paired/" }, "v13_native": { "cases_passed": 6, "cases_total": 6, "listing_clients": 45, "booked_snapshot_total_gbp": "2454.50", "carried_count": 45, "final_reads": 30, "new_executions_on_final_reads": 0, "ready_to_checked_dom_ms": [ 390.51, 220.6 ], "optional_delay_seconds": 20, "full_surfaces": "table, CSV, share, reload", "duplicate_grain_refusal": "actual separate fixture with repeated user_id", "evidence": "results-v8/structural-browser-v13-results.json", "scope": "External generation intercepted; native API/worker/PG/persistence/rendering" }, "v13_development": { "candidate_revision": "83406abb20b49dd13cfa7e8fa144bdb555c85b5d", "attempts_per_arm": 70, "first_turns": 28, "candidate_correct_first_turns": 8, "candidate_first_turn_errors": 4, "candidate_precision": "8/12", "candidate_silent_wrong": 4, "candidate_conversation_success": "0/14", "claim_scope": "Opened development only. No independent claim. All original scores retained.", "evidence": "phase0/development-r2-v13-paired/paired-summary.json" }, "v14_pre_freeze": { "source_manifest_sha256": "5698f24ebd6522e53279e298dc949eb26bc54cc43c65906011892a91ba27dca0", "source_files": 1692, "query_gate": { "passed": 268, "failed": 0, "skipped": 0 }, "root_calendar_gate": { "passed": 98, "failed": 0, "skipped": 0 }, "root_route_neighbors": { "passed": 68, "failed": 0 }, "result_gate": { "passed": 135, "failed": 0 }, "independent_result_final_gate": { "passed": 22, "failed": 0, "skipped": 0 }, "counts_overlap": true, "reviewers": "Three requested gpt-6-astra/xhigh lanes; separate ownership and cross-contract reviews", "open_concrete_review_blockers": 0, "development_revision": "r3; seven disproved references reconciled; all70IDs and all old scores retained", "final_native_full_suite_latency_holdout": "pending" }, "working_candidate": null, "v14_development": { "attempts": 70, "correct_completed": 25, "silent_wrong": 2, "comparison_unverified": 1, "refusal_unverified": 17, "justified_refusal": 2, "blocked_by_parent": 16, "task_error": 5, "unavailable_exports": 2, "first_turn_errors": 2, "first_turn_correct": 10, "first_turn_denominator": 28, "precision_denominator": 27, "conversations_successful": 0, "conversations": 14, "deadline_jobs": 0, "jobs": 67, "narrow_paired_regressions": 0, "numerically_correct_with_extra_predicate": 3, "quality_claim_scope": "Opened development only." }, "v14_full_suite": { "complete": false, "tool_exit_code": 143, "last_progress_percent": 99, "final_counts": "unavailable", "source_and_test_hash_changes": 0, "evidence": "tests-v14/full-suite-interrupted/interruption.json" }, "v15_development": { "attempts": 70, "correct_completed": 24, "silent_wrong": 2, "comparison_unverified": 2, "refusal_unverified": 18, "justified_refusal": 2, "blocked_by_parent": 15, "task_error": 5, "unavailable_csv": 2, "first_turn_correct": 10, "first_turn_errors": 2, "first_turn_denominator": 28, "full_conversations": 0, "conversation_denominator": 14, "unnecessary_clarifications": 0, "successful_parent_denominator": 12, "jobs": 67, "deadline_jobs": 0, "quality_certification": false, "evidence": "phase0/development-r3-v15-paired/" }, "v15_full_suite": { "passed": 29708, "failed": 23, "skipped": 50, "seconds": 711.29, "workers": 8, "source_test_mismatches": 0, "failures": { "preexisting": 21, "stale_test_contract": 1, "product": 1 }, "evidence": "tests-v15/full-suite/" }, "v16_scoped": { "application_source_manifest_sha256": "d3da27a645c2300f365fce10b7adfdec2cd362ae6c7331668ff126da15735cd4", "query_source": { "passed": 154, "skipped": 0 }, "carry_routing_native": { "passed": 50, "skipped": 0 }, "provider_and_neighbors": { "passed": 67, "skipped": 0 }, "calendar": { "passed": 125, "skipped": 0 }, "booking_route": { "passed": 193, "skipped": 0 }, "result_python": { "passed": 133 }, "node": { "passed": 244 }, "chromium_mocked": { "passed": 77 }, "note": "Gates overlap. Counts are not additive. Full frozen/native/live measurements remain pending." }, "historical_metadata_before_v17": { "application_manifest_sha256": "4888549e416c140680e8f95966d0df6338d57c3248d46eab9dcf59fef09a49c0", "source_scope": "Historical V7 evidence is retained. E5 environment is accepted under the user amendment. E6 v4 is frozen and its API boundary preflight passed. V9 development is complete; V10 fixes are in progress. Final candidate holdout remains unadmitted.", "active_checkpoint": "49914705299ae3be3ecf59a07ba8602370883539", "application_checkpoint": "49914705299ae3be3ecf59a07ba8602370883539", "application_manifest_scope": "Exact V9 development application snapshot. Subsequent V10 working changes require a new freeze and verification.", "active_candidate_checkpoint": "49914705299ae3be3ecf59a07ba8602370883539", "working_candidate": "V16 source stable; final named checkpoint and evaluation pending" }, "v16_full_suite": { "passed": 29817, "failed": 21, "skipped": 50, "seconds": 715.67, "workers": 8, "source_test_mismatches": 0, "preexisting_failures": 21, "green": false, "evidence": "tests-v16/full-suite/" }, "v16_development": { "attempts": 70, "correct_completed": 22, "silent_wrong": 2, "comparison_unverified": 4, "refusal_unverified": 21, "justified_refusal": 2, "blocked_by_parent": 16, "task_error": 1, "unavailable_csv": 2, "first_turn_correct": 6, "first_turn_errors": 1, "first_turn_denominator": 28, "precision": "22/24", "full_conversations": 0, "conversation_denominator": 14, "unnecessary_clarifications": 0, "successful_parent_denominator": 10, "jobs": 67, "deadline_jobs": 0, "correct_to_error_or_wrong": 0, "quality_certification": false, "evidence": "phase0/development-r3-v16-paired/" }, "v16_native": { "cases_passed": 6, "cases": 6, "listing_members": 45, "snapshot_gbp": "2454.50", "carried_count": 45, "final_reads": 30, "new_work_on_final_reads": 0, "ready_to_dom_ms": [ 285.054, 289.813 ], "optional_delays_seconds": 20, "latency_scope": "Two native browser observations during concurrent work; not E6 p95.", "evidence": "results-v8/structural-browser-v16-results.json" }, "v17_working_checks": { "rank_thresholds": { "passed": 165, "skipped": 0 }, "membership_predicate_boundaries": { "passed": 47, "skipped": 0 }, "distinct_snapshot_sum_broader": { "passed": 270, "skipped": 0 }, "explicit_output_unit_initial": { "passed": 19 }, "counts_overlap": true, "integrated_candidate": { "passed": 701, "failed": 0, "skipped": 0, "input_changes": 0, "scope": "Pre-freeze scoped code" }, "reviewers": "All three requested gpt-6-astra/xhigh agents stopped at service usage limit. Lead continues coordinated files. No model substitution.", "mypy": "3 changed sources pass strict", "ruff": "All changed/new source and test files pass", "evidence": "tests-v17/scoped/receipt.json" }, "v17_full_suite": { "passed": 29898, "failed": 21, "skipped": 50, "seconds": 727.85, "workers": 8, "source_test_mismatches": 0, "preexisting_failures": 21, "green": false, "evidence": "tests-v17/full-suite/" }, "v17_development": { "attempts": 70, "correct_completed": 20, "silent_wrong": 0, "comparison_unverified": 0, "refusal_unverified": 23, "justified_refusal": 2, "blocked_by_parent": 20, "task_error": 3, "unavailable_csv": 2, "first_turn_correct": 7, "first_turn_errors": 1, "first_turn_denominator": 28, "precision": "20/20", "full_conversations": 0, "conversation_denominator": 14, "unnecessary_clarifications": 0, "successful_parent_denominator": 4, "jobs": 67, "deadline_jobs": 0, "correct_to_error_or_wrong": 0, "quality_certification": false, "evidence": "phase0/development-r3-v17-paired/" }, "v17_native": { "cases_passed": 6, "cases": 6, "listing_members": 45, "snapshot_gbp": "2454.50", "carried_count": 45, "final_reads": 30, "new_work_on_final_reads": 0, "ready_to_dom_ms": [ 252.676251, 315.1177 ], "optional_delays_seconds": 20, "latency_scope": "Two native browser observations during concurrent work; not E6 p95.", "evidence": "results-v8/structural-browser-v17-results.json" }, "history_before_final_v18_freeze": { "application_manifest_sha256": "3095200d047a9f3310fdb8f4e211119ff7a6d829170b4574f81495335e4edb60", "final_quality_status": "Unverified: final candidate, isolated E6 and independent holdout are unadmitted. Requested reviewers unavailable at service usage limit.", "active_checkpoint": "c85632b1a80e6517f3359531096b0f9496ebdec3", "application_checkpoint": "c85632b1a80e6517f3359531096b0f9496ebdec3", "active_candidate_checkpoint": "c85632b1a80e6517f3359531096b0f9496ebdec3", "working_candidate": "V18 working membership output and money correction boundaries; not frozen or independently reviewed." }, "v18_full_suite": { "candidate_revision": "8958ae93ff196bdc5fbcfd08ae12f81445c5d28c", "full_suite": { "passed": 29927, "failed": 21, "skipped": 50, "workers": 8, "application_or_test_changes": 0, "junit_seconds": 727.123, "pytest_reported_seconds": 731.98 }, "all_failures_match_exact_v17_nodes": true, "v17_original_classification": "tests-v17/full-suite/failure-disposition.json", "v17_classification_sha256": "e8ad7be0efd42703e79a117b4dcbf2b61344e0e5a14b07f3eb23cdadcb91e6f8", "new_failed_nodes": [], "green_full_suite": false, "cases": [ "tests.test_bench_adapters::test_legacy_adapters_cover_all_five_eval_collections_with_verified_counts", "tests.test_bench_adapters::test_load_cases_merges_legacy_adapters_and_yaml_with_unique_stable_ids", "tests.test_bi_paraphrase_eval_cli::test_default_corpus_covers_all_required_golden_keys", "tests.test_c2_run_rate_sar_semantics::test_non_ct_questions_do_not_escalate[how is onboarding performing this month]", "tests.test_c2_run_rate_sar_semantics::test_non_ct_questions_do_not_escalate[new signups by month]", "tests.test_compose_floor_preflight::test_all_clear_proceeds_and_drafts", "tests.test_compose_floor_preflight::test_missing_partner_client_pain_blocks_partner_affiliate", "tests.test_compose_floor_preflight::test_needs_human_floor_fact_blocks_naming_the_slot", "tests.test_compose_floor_preflight::test_non_floor_zero_count_slot_warns_but_does_not_block", "tests.test_compose_floor_preflight::test_none_in_floor_slot_blocks", "tests.test_compose_floor_preflight::test_one_pager_manifest_enumerates_the_exact_floor", "tests.test_compose_floor_preflight::test_three_failures_aggregate_into_one_422_with_zero_drafts", "tests.test_compose_floor_preflight::test_too_few_pillars_blocks_as_canon_fact", "tests.test_compose_floor_preflight::test_zero_chunks_on_floor_proof_slot_blocks", "tests.test_content_deck_templates_b.TestFloorManifest::test_manifest_surfaces_on_the_templates_api", "tests.test_content_deck_templates_b.TestFloorManifest::test_one_pager_manifest_enumerates_the_exact_floor", "tests.test_content_deck_templates_b.TestFloorManifest::test_partner_manifest_declares_the_five_required_brief_fields", "tests.test_eval_oracle_playwright_harness::test_eval_gate2b_differential_scorer_detects_selection_ignored", "tests.test_eval_oracle_playwright_harness::test_eval_gate2b_quarantined_combo_does_not_block_remaining_browser_track", "tests.test_span_refine_marker_splice::test_parser_leaves_plain_numbered_items_alone", "tests.test_span_refine_marker_splice::test_refine_span_founder_scenario_stores_single_marker" ] }, "v18_native": { "candidate_revision": "8958ae93ff196bdc5fbcfd08ae12f81445c5d28c", "source_manifest_sha256": "39a7ef2f2a147ce584606a711055de14193a5a226a7f5ff387ad6d8fdabe0165", "evidence_kind": "Native Chromium -> API -> worker -> disposable PostgreSQL -> SQLite -> rendered answer; only external generation intercepted", "registered_case_gate": { "passed": 6, "failed": 0, "attempts": 2, "submitted_case_runs": 1, "startup_navigation_failures": 1 }, "checked_success_facts": { "listing_members": 45, "scoped_total_gbp": "2454.50", "checked_membership_count": 45, "surface_modes": [ "saved API", "table", "headline", "snapshot CSV", "authorized same-vault full-details share API", "share DOM", "share reload" ] }, "final_reads": { "attempted": 30, "new_source_execution": 0, "new_provider_generation": 0, "new_jobs": 0, "new_usage_rows": 0 }, "optional_readiness": [ { "case_id": "listing", "required_ready_to_visible_ms": 225.49805510789156, "actual_optional_delay_ms": 20000.490977428854, "readable_before_optional_completion": true, "native_monotonic_clock_verified": true, "job_correlation": { "source": "native SQLite bi_turn_jobs", "turn_id": "1a966596-6556-40bf-aeea-96158ce1957b", "job_id": "ae1b940f-c186-442e-8e3b-c9c3ebb1c0cd", "matched_rows": 1, "raw_readiness_job_id": "ae1b940f-c186-442e-8e3b-c9c3ebb1c0cd" } }, { "case_id": "scoped_total", "required_ready_to_visible_ms": 334.19828303158283, "actual_optional_delay_ms": 20004.91074193269, "readable_before_optional_completion": true, "native_monotonic_clock_verified": true, "job_correlation": { "source": "native SQLite bi_turn_jobs", "turn_id": "d9a27cf1-2e34-4281-95d2-1b117ed1537c", "job_id": "371db928-cd89-4b65-ad41-3612d1de6a16", "matched_rows": 1, "raw_readiness_job_id": "371db928-cd89-4b65-ad41-3612d1de6a16" } } ], "safe_refusal_outcome_parity": { "passed": 2, "failed": 0, "classes": [ "permission", "grain" ], "top_level_answer_record_confidence_record_agree": true, "read_only_receipt": "native-browser-v18/refusal-parity.json" }, "footer": { "status": "passed", "no_false_checked_rows_footer_all_six": true }, "jobs": { "done": 6, "deadline": 0, "active_at_stop": 0 }, "console_errors": 0, "http_errors": 0, "external_network_guard_violations": 0, "live_model_quality_claim": false, "holdout_quality_claim": false, "deterministic_latency_claim": false, "limits": [ "External generation is declared and intercepted; HTTP, API, worker, PostgreSQL, persistence and rendered answer are native. This is not a live-model quality run.", "The three-turn successful chain and two safe refusals are six public development cases. Five-turn failure-position tests are separate deterministic controls.", "Readiness includes IPC, layout and two animation frames while full Python and live-model work run concurrently. These observations are not pixel paint, p95 or E6.", "Native money is GBP. Mixed currencies, empty-result recovery and other interface states have separate deterministic or mocked-browser coverage. Native Compose and refreshed CSV after data changes are not covered.", "The lead performed this source binding and visual review. The three requested independent reviewers are unavailable at their model service limit." ] }, "v18_e6": { "timing_purpose": "uncontended_acceptance", "paired_configuration": { "equal": true, "excluded_replica_identity_keys": [ "LORE_BI_FEDERATION_CT_CONNECTION_ID", "LORE_BI_FEDERATION_GOLD_CONNECTION_ID", "LORE_BI_FEDERATION_VAULT_ID", "LORE_DB_PATH" ], "shared_configuration_sha256": "58fc02223a1f8d38dec9a5e0726dd70850c138b4c2e55d76cfff496ae25407aa" }, "workload_sha256": "193aed08832bb0feb808c0e0c3bd4817b7e09a8b3ff7a13a140f46aa46ad963e", "response_supplement_sha256": "56e8a72041ecf76af6324fd7f0f1e9b0ca7ef430d50ca79ace464513aa7789eb", "arms": { "original": { "attempts": 90, "p95_ms": 1549.4488747790456, "successful_results": 54, "completed_fixed_queries": 36, "terminal_errors_or_refusals": 36, "fixed_provider_proven_for_every_job": true, "all_external_calls": { "valid": true, "failure_codes": [], "sql_external_calls": 159, "auxiliary_external_calls": 87, "persisted_call_records": 246, "usage_ledger_calls": 87, "guard_violations": 0 } }, "candidate": { "attempts": 90, "p95_ms": 1732.0084376260638, "successful_results": 69, "completed_fixed_queries": 66, "terminal_errors_or_refusals": 21, "fixed_provider_proven_for_every_job": true, "all_external_calls": { "valid": true, "failure_codes": [], "sql_external_calls": 111, "auxiliary_external_calls": 69, "persisted_call_records": 180, "usage_ledger_calls": 69, "guard_violations": 0 } } }, "candidate_to_original_p95_ratio": 1.1178222565575464, "all_fixed_queries_completed": false, "target": "failed", "numerical_p95_ratio_at_most_1_10": false, "fixed_provider_verified": true, "non_answers_without_persisted_llm_calls": 57, "missing_job_context_association": "Keep raw null job_id. Associate fixed case and complete provider interval within the client's real-monotonic attempt window.", "workload_conformity_checked_outside_timing": true, "request_row_limit": 10000, "enrichment_policy": "Only additive client/client_type columns beside a requested user_id. All requested fields, order, membership, values and types remain exact. No truncated result completes the fixed query.", "poll_interval_ms": 250, "timing_scope": "Client monotonic submission through terminal persisted result read. Includes 250 ms polling resolution and HTTP overhead; not worker-only execution time.", "all_timings_retained": true, "timings_filtered_for_correctness": false, "no_quality_claim_from_deterministic_provider": true }, "independent_final_review": "All three requested gpt-6-astra/xhigh reviewers completed the final challenge. They found material unresolved defects. This is separate-agent review, not an external audit." }