Record decision trace eval ownership boundary

README.md

@@ -428,6 +428,7 @@ For a guided walkthrough of the canonical demo flow, see [docs/demo/fog-harbor-w
 - Post-Phase-67 outcome/report/eval generalization audit keeps the queue paused; `docs/plans/post-phase-67-outcome-report-eval-generalization-audit-2026-06-08.md` records that selected-world proof is current for `fog-harbor-east-gate`, `museum-night`, and `library-rain`, does not claim future-world readiness, and does not prove a Phase 68 blocker. Do not open Phase 68 from this audit.
 - Post-Phase-67 runtime-created world eval proof keeps the queue paused; `docs/plans/post-phase-67-runtime-created-world-eval-proof-2026-06-08.md` records that a runtime-created bounded world can pass `eval-world`, does not claim future-world readiness, and does not prove a Phase 68 blocker. Do not open Phase 68 from this proof.
 - Post-Phase-67 decision-trace replay audit keeps the queue paused; `docs/plans/post-phase-67-decision-trace-replay-audit-2026-06-08.md` records current evidence for kernel-level replay and runner-level same-run-directory replay. Do not open Phase 68 from this audit.
+- Post-Phase-67 decision-trace eval-ownership boundary keeps the queue paused; `docs/plans/post-phase-67-decision-trace-eval-ownership-boundary-2026-06-08.md` records that eval-owned decision-trace replay metrics remain unclaimed and that opening implementation without a source-backed blocker would be blueprint drift. Do not open Phase 68 from this boundary note.
 - Phase 48 is closed after PR `#382`, issue `#375`, and milestone `Phase 48 - Successor Intake and Boundary Contract Triage`.
 - Phase 49 is closed after PR `#395`, issue `#383`, and milestone `Phase 49 - Kernel, Perturbation, and Runtime Contract Hardening`; completed work items were `#384`, `#386`, `#388`, `#390`, `#392`, and `#394`.
 - Phase 50 is closed after PR `#402`, issue `#396`, and milestone `Phase 50 - Runtime Orchestration Measurement and Product Boundary`; completed work items were `#397`, `#398`, and `#401`. Phase 50 measured before any `task_id` or worker contract is introduced and kept the private-beta launch hub planning-only for now.

backend/tests/test_post_phase67_decision_trace_eval_ownership_boundary.py

@@ -0,0 +1,122 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+
+BOUNDARY_PATH = Path("docs/plans/post-phase-67-decision-trace-eval-ownership-boundary-2026-06-08.md")
+REPLAY_AUDIT_PATH = Path("docs/plans/post-phase-67-decision-trace-replay-audit-2026-06-08.md")
+MINIMUM_LOOP = (
+    "corpus -> chunks -> graph -> personas -> scenarios -> deterministic runs -> "
+    "report/claims -> eval"
+)
+
+
+def _read(path: Path) -> str:
+    assert path.exists(), path
+    return path.read_text(encoding="utf-8")
+
+
+def test_decision_trace_eval_ownership_boundary_exists_with_required_sections() -> None:
+    boundary = _read(BOUNDARY_PATH)
+    required_sections = [
+        "# Post-Phase-67 Decision Trace Eval Ownership Boundary",
+        "## Scope",
+        "## Current-Code Finding",
+        "## Source Evidence",
+        "## Blueprint Alignment Decision",
+        "## Future Trigger Conditions",
+        "## Boundaries",
+        "## Validation Commands",
+    ]
+    for section in required_sections:
+        assert section in boundary
+
+
+def test_eval_ownership_boundary_keeps_metrics_unclaimed_without_phase68() -> None:
+    boundary = _read(BOUNDARY_PATH)
+    replay_audit = _read(REPLAY_AUDIT_PATH)
+    contracts = _read(Path("docs/architecture/contracts.md"))
+    eval_service = _read(Path("backend/app/evals/service.py"))
+    decision_kernel = _read(Path("backend/app/decision_kernel/service.py"))
+    pipeline_tests = _read(Path("backend/tests/test_pipeline.py"))
+
+    required_boundary_phrases = [
+        f"`{MINIMUM_LOOP}`",
+        "Current-code audit result: eval-owned decision-trace replay metrics remain unclaimed today.",
+        "This is not a source-backed Phase 68 blocker.",
+        "Opening an implementation queue for eval-owned decision-trace replay metrics without a new source-backed blocker would be blueprint drift.",
+        "Do not open Phase 68 from this boundary note.",
+        "The current queue remains in the formal paused stop-state.",
+        "`backend/app/evals/service.py` currently has no `decision_trace`, `replay_cache`, or `accepted_from_replay` ownership.",
+        "`docs/architecture/contracts.md` does not require transfer eval summaries to include decision-trace replay metrics.",
+        "kernel-level replay and runner-level same-run-directory replay remain the current proven boundary",
+        "No ADR or `docs/architecture/contracts.md` update is made by this boundary note because this diff does not change a protected-core contract.",
+        "`status:needs-adr` and unresolved `risk:safety` findings remain merge blockers.",
+        "Do not present Mirror as a real-world prediction machine.",
+        "Do not build real-person personas or digital doubles.",
+        "Do not build political persuasion, hidden surveillance, law-enforcement scoring, hiring, credit, medical, or judicial decision systems.",
+    ]
+    forbidden_boundary_phrases = [
+        "Phase 68 is active",
+        "Phase 68 execution queue is open",
+        "open a Phase 68 milestone now",
+        "eval-owned decision-trace replay metrics are implemented",
+        "changes decision_trace.jsonl shape",
+        "claims provider-backed replay readiness",
+        "claims future-world readiness",
+    ]
+    for phrase in required_boundary_phrases:
+        assert phrase in boundary, phrase
+    for phrase in forbidden_boundary_phrases:
+        assert phrase not in boundary, phrase
+
+    assert "Future eval ownership for decision-trace replay metrics remains unclaimed by this audit." not in replay_audit
+    assert "`docs/plans/post-phase-67-decision-trace-eval-ownership-boundary-2026-06-08.md`" in replay_audit
+
+    for phrase in [
+        "`decision_trace.jsonl` is the durable v1 decision audit artifact",
+        "`replay_cache` when the selection is copied from an existing trace entry for the same",
+        "`accepted_from_replay` for replay-cache reuse.",
+    ]:
+        assert phrase in contracts
+
+    for phrase in ["decision_trace", "replay_cache", "accepted_from_replay"]:
+        assert phrase not in eval_service
+
+    for phrase in [
+        "replay_entry = self.replay_cache.get(input_hash)",
+        'provider_mode="replay_cache"',
+        'validation_status="accepted_from_replay"',
+    ]:
+        assert phrase in decision_kernel
+        assert phrase in boundary
+
+    assert "test_simulation_replays_from_existing_decision_trace" in pipeline_tests
+    assert "test_simulation_replays_from_existing_decision_trace" in boundary
+
+
+def test_current_docs_point_to_eval_ownership_boundary_without_opening_phase68() -> None:
+    docs = [
+        Path("README.md"),
+        Path("docs/plans/current-state-baseline.md"),
+        Path("docs/plans/phase-execution-queue.md"),
+        Path("docs/plans/automation-roadmap.md"),
+    ]
+    required_phrases = [
+        "`docs/plans/post-phase-67-decision-trace-eval-ownership-boundary-2026-06-08.md`",
+        "Post-Phase-67 decision-trace eval-ownership boundary keeps the queue paused",
+        "eval-owned decision-trace replay metrics remain unclaimed",
+        "Do not open Phase 68 from this boundary note.",
+    ]
+    forbidden_phrases = [
+        "Phase 68 is active",
+        "Phase 68 execution queue is open",
+        "`audit-github-queue` reports `ready` for Phase 68",
+        "milestone `Phase 68",
+    ]
+    for path in docs:
+        text = _read(path)
+        for phrase in required_phrases:
+            assert phrase in text, f"{path} is missing eval-ownership boundary pointer: {phrase}"
+        for phrase in forbidden_phrases:
+            assert phrase not in text, f"{path} opens Phase 68 prematurely: {phrase}"

backend/tests/test_post_phase67_decision_trace_replay_audit.py

@@ -45,7 +45,7 @@ def test_decision_trace_replay_audit_is_source_backed_without_phase68() -> None:
         "The current queue remains in the formal paused stop-state.",
         "does not change `decision_trace.jsonl` shape",
         "No ADR or `docs/architecture/contracts.md` update is made by this audit because this diff does not change a protected-core contract.",
-        "TODO[verify]: Future eval ownership for decision-trace replay metrics remains unclaimed by this audit.",
+        "`docs/plans/post-phase-67-decision-trace-eval-ownership-boundary-2026-06-08.md` records that future eval ownership for decision-trace replay metrics remains unclaimed and is not a current Phase 68 blocker.",
         "`status:needs-adr` and unresolved `risk:safety` findings remain merge blockers.",
         "Do not present Mirror as a real-world prediction machine.",
         "Do not build real-person personas or digital doubles.",

docs/plans/automation-roadmap.md

@@ -31,6 +31,7 @@ Day 0 bootstrap is complete, Phase 5 closeout is complete, Phase 6 closeout is c
 - Post-Phase-67 outcome/report/eval generalization audit keeps the queue paused; `docs/plans/post-phase-67-outcome-report-eval-generalization-audit-2026-06-08.md` records that selected-world proof is current for `fog-harbor-east-gate`, `museum-night`, and `library-rain`, does not claim future-world readiness, and does not prove a Phase 68 blocker. Do not open Phase 68 from this audit.
 - Post-Phase-67 runtime-created world eval proof keeps the queue paused; `docs/plans/post-phase-67-runtime-created-world-eval-proof-2026-06-08.md` records that a runtime-created bounded world can pass `eval-world`, does not claim future-world readiness, and does not prove a Phase 68 blocker. Do not open Phase 68 from this proof.
 - Post-Phase-67 decision-trace replay audit keeps the queue paused; `docs/plans/post-phase-67-decision-trace-replay-audit-2026-06-08.md` records current evidence for kernel-level replay and runner-level same-run-directory replay. Do not open Phase 68 from this audit.
+- Post-Phase-67 decision-trace eval-ownership boundary keeps the queue paused; `docs/plans/post-phase-67-decision-trace-eval-ownership-boundary-2026-06-08.md` records that eval-owned decision-trace replay metrics remain unclaimed and that opening implementation without a source-backed blocker would be blueprint drift. Do not open Phase 68 from this boundary note.
 - Phase 1 and Phase 2 gates are closed.
 - Phase 3 is closed locally and in GitHub.
 - Phase 3 exit issue `#4` is closed and milestone `Phase 3 - Eval/UI/Demo` is closed.

docs/plans/current-state-baseline.md

@@ -32,6 +32,8 @@ Post-Phase-67 runtime-created world eval proof keeps the queue paused; `docs/pla
 
 Post-Phase-67 decision-trace replay audit keeps the queue paused; `docs/plans/post-phase-67-decision-trace-replay-audit-2026-06-08.md` records current evidence for kernel-level replay and runner-level same-run-directory replay. Do not open Phase 68 from this audit.
 
+Post-Phase-67 decision-trace eval-ownership boundary keeps the queue paused; `docs/plans/post-phase-67-decision-trace-eval-ownership-boundary-2026-06-08.md` records that eval-owned decision-trace replay metrics remain unclaimed and that opening implementation without a source-backed blocker would be blueprint drift. Do not open Phase 68 from this boundary note.
+
 ## Snapshot
 
 - Local quality baseline:

docs/plans/phase-execution-queue.md

@@ -120,6 +120,7 @@ Phase 67 - Blueprint Calibration and Minimum-Loop Value Gate
   - Post-Phase-67 outcome/report/eval generalization audit keeps the queue paused; `docs/plans/post-phase-67-outcome-report-eval-generalization-audit-2026-06-08.md` records that selected-world proof is current for `fog-harbor-east-gate`, `museum-night`, and `library-rain`, does not claim future-world readiness, and does not prove a Phase 68 blocker. Do not open Phase 68 from this audit.
   - Post-Phase-67 runtime-created world eval proof keeps the queue paused; `docs/plans/post-phase-67-runtime-created-world-eval-proof-2026-06-08.md` records that a runtime-created bounded world can pass `eval-world`, does not claim future-world readiness, and does not prove a Phase 68 blocker. Do not open Phase 68 from this proof.
   - Post-Phase-67 decision-trace replay audit keeps the queue paused; `docs/plans/post-phase-67-decision-trace-replay-audit-2026-06-08.md` records current evidence for kernel-level replay and runner-level same-run-directory replay. Do not open Phase 68 from this audit.
+  - Post-Phase-67 decision-trace eval-ownership boundary keeps the queue paused; `docs/plans/post-phase-67-decision-trace-eval-ownership-boundary-2026-06-08.md` records that eval-owned decision-trace replay metrics remain unclaimed and that opening implementation without a source-backed blocker would be blueprint drift. Do not open Phase 68 from this boundary note.
   - If a future minimum-loop audit finds a contract gap, split a separate protected-core contract issue before changing schema, scenario DSL, claim labels, run trace shape, or artifact layout.
 
 ## Phase 66 Closed Queue

docs/plans/post-phase-67-decision-trace-eval-ownership-boundary-2026-06-08.md

@@ -0,0 +1,77 @@
+# Post-Phase-67 Decision Trace Eval Ownership Boundary
+
+Date: 2026-06-08
+
+## Scope
+
+This boundary note closes the eval-ownership follow-up left by the Post-Phase-67 decision-trace replay audit.
+
+It asks only whether future eval ownership for decision-trace replay metrics is a current source-backed blocker. It does not implement new eval metrics.
+
+The minimum loop remains:
+
+```text
+corpus -> chunks -> graph -> personas -> scenarios -> deterministic runs -> report/claims -> eval
+```
+
+The queue shorthand is `corpus -> chunks -> graph -> personas -> scenarios -> deterministic runs -> report/claims -> eval`.
+
+## Current-Code Finding
+
+Current-code audit result: eval-owned decision-trace replay metrics remain unclaimed today.
+
+This is not a source-backed Phase 68 blocker.
+
+The current queue remains in the formal paused stop-state.
+
+`backend/app/evals/service.py` currently has no `decision_trace`, `replay_cache`, or `accepted_from_replay` ownership.
+
+`docs/architecture/contracts.md` does not require transfer eval summaries to include decision-trace replay metrics.
+
+## Source Evidence
+
+- `docs/architecture/contracts.md` states that `decision_trace.jsonl` is the durable v1 decision audit artifact.
+- `docs/architecture/contracts.md` defines `replay_cache` when the selection is copied from an existing trace entry for the same `input_hash`.
+- `docs/architecture/contracts.md` defines `accepted_from_replay` for replay-cache reuse.
+- `backend/app/decision_kernel/service.py` looks up `replay_entry = self.replay_cache.get(input_hash)`.
+- `backend/app/decision_kernel/service.py` finalizes cache hits with `provider_mode="replay_cache"` and `validation_status="accepted_from_replay"`.
+- `backend/tests/test_pipeline.py` includes `test_simulation_replays_from_existing_decision_trace`.
+
+## Blueprint Alignment Decision
+
+Do not open Phase 68 from this boundary note.
+
+Opening an implementation queue for eval-owned decision-trace replay metrics without a new source-backed blocker would be blueprint drift.
+
+The correct current boundary is narrower: kernel-level replay and runner-level same-run-directory replay remain the current proven boundary, while eval-owned decision-trace replay metrics remain future contract work only if a later audit proves a minimum-loop gap.
+
+No successor queue is opened by this boundary note.
+
+## Future Trigger Conditions
+
+Before eval-owned decision-trace replay metrics can become implementation work, a future reviewed audit must identify:
+
+- the exact metric eval would own, such as trace existence, replay consistency, cache-hit rate, provider-call avoidance, privacy redaction, or runtime-node coverage
+- the eval command and artifact scope, such as `eval-world`, `eval-transfer`, generated runtime sessions, or private-beta-only validation
+- the source-backed minimum-loop gap or protected-core contract blocker that makes the metric necessary now
+- whether `eval/summary.json` becomes a stable metric contract and therefore needs `docs/architecture/contracts.md` and possibly an ADR
+
+## Boundaries
+
+- This boundary note does not change `decision_trace.jsonl` shape.
+- This boundary note does not change schema, scenario DSL, perturbation payload schema, decision schema, claim labels, report claim `evidence_ids`, run trace shape, compare artifact shape, session/node manifest shape, public demo artifact layout, plugin MCP contract, route ownership, or artifact layout.
+- This boundary note does not assert provider-backed replay readiness.
+- This boundary note does not assert future-world readiness.
+- No ADR or `docs/architecture/contracts.md` update is made by this boundary note because this diff does not change a protected-core contract.
+- `status:needs-adr` and unresolved `risk:safety` findings remain merge blockers.
+- Do not present Mirror as a real-world prediction machine.
+- Do not build real-person personas or digital doubles.
+- Do not build political persuasion, hidden surveillance, law-enforcement scoring, hiring, credit, medical, or judicial decision systems.
+
+## Validation Commands
+
+- `python -m pytest backend/tests/test_post_phase67_decision_trace_eval_ownership_boundary.py -q`
+- `python -m pytest backend/tests/test_post_phase67_decision_trace_replay_audit.py backend/tests/test_post_phase67_decision_trace_eval_ownership_boundary.py -q`
+- `python scripts/check_no_secrets.py`
+- `python -m backend.app.cli audit-github-queue --repo YSCJRH/mirror-sim`
+- `git diff --check`

docs/plans/post-phase-67-decision-trace-replay-audit-2026-06-08.md

@@ -47,7 +47,7 @@ This audit does not prove a protected-core blocker. It proves that the current d
 
 No successor queue is opened by this audit.
 
-TODO[verify]: Future eval ownership for decision-trace replay metrics remains unclaimed by this audit.
+Follow-up boundary: `docs/plans/post-phase-67-decision-trace-eval-ownership-boundary-2026-06-08.md` records that future eval ownership for decision-trace replay metrics remains unclaimed and is not a current Phase 68 blocker.
 
 ## Boundaries