fix(loop-memory): isolate cross-initiative authority evidence

.agent-loop/initiatives/WS-ENG-007-concurrent-pr-review-reconciliation/CHUNK_MAP.md

@@ -4,12 +4,14 @@
 |---:|---|---|---:|---|
 | 0 | `WS-ENG-007-00R1` | Repair planning-intake file/tree parity and recover PR #187 exactly once | L1/P0 | Merged; recovery superseded after rerun-cardinality failure |
 | 0 | `WS-ENG-007-00R2` | Canonicalize repeated trusted check evidence and reconcile PRs #187 and #188 exactly once | L1/P0 | Merged; superseded by mutable-history failure |
-| 0 | `WS-ENG-007-00R3` | Freeze accepted checks at merge time and give explicit starts deterministic recovery parity | L1/P0 | Active fail-closed recovery |
-| 1 | `WS-ENG-007-01` | Add deterministic reviewed-patch identity and conservative base-delta review preservation | L1 | Blocked on 00R3 merge, successful four-merge reconciliation, and explicit start |
+| 0 | `WS-ENG-007-00R3` | Freeze accepted checks at merge time and give explicit starts deterministic recovery parity | L1/P0 | Merged; exposed cross-initiative projection mixing |
+| 0 | `WS-ENG-007-00R4` | Separate global merge evidence from initiative-local authority projections | L1/P0 | Active fail-closed recovery after the first cross-initiative start exposed projection mixing |
+| 1 | `WS-ENG-007-01` | Add deterministic reviewed-patch identity and conservative base-delta review preservation | L1 | Blocked on 00R4 merge, successful reconciliation, and explicit start |
 | 2 | `WS-ENG-007-02` | Add structured reviewer-track and upstream-finding reconciliation | L1 | Blocked on 01 merge and explicit start |
 | 3 | `WS-ENG-007-03` | Add merge-group CI parity and queue-readiness proof | L1 | Blocked on 02 merge and explicit start |
 
-Recovery chunks are exceptional ordered prerequisites; `00R3` supersedes the
-incomplete mutable-evidence behavior left after `00R2`. Implementation chunks
+Recovery chunks are exceptional ordered prerequisites; `00R4` preserves the
+merge-bound evidence fixed by `00R3` while repairing cross-initiative authority
+projection composition. Implementation chunks
 remain one PR each and stop after merge. Every successor requires a separate
 explicit signed start.

.agent-loop/initiatives/WS-ENG-007-concurrent-pr-review-reconciliation/STATUS.md

@@ -1,18 +1,19 @@
 # STATUS: WS-ENG-007 - Concurrent PR Review Reconciliation
 
-- Phase: merge-bound check-evidence recovery
+- Phase: cross-initiative authority-projection recovery
 - Gate: fail-closed automation repair
 - Active planning chunk: none
 - Active implementation chunk: none
 - Completed but unreconciled recovery chunk: `WS-ENG-007-00R1`
 - Merged but unreconciled recovery chunk: `WS-ENG-007-00R2`
-- Active recovery chunk: `WS-ENG-007-00R3`
+- Completed recovery chunk: `WS-ENG-007-00R3`
+- Active recovery chunk: `WS-ENG-007-00R4`
 - Proposed implementation successor after recovery: `WS-ENG-007-01`
 - Separate explicit start required: true
-- Current gate: PRs #187–#189 are merged, but signed reconciliation still fails
-  because mutable post-merge check history is re-evaluated and the explicit
-  start workflow lacks the merge workflow's recovery path. Signed state remains
-  at `73b457925b02301587b83d01ced0adb66319d134`; no successor is active.
-- Review gate: all required internal tracks passed reviewed implementation
-  `4a3eee7fdd80970675ae2574542073b8a92e1ef9`; awaiting external checks and the
+- Current gate: PR #190 reconciled signed memory through
+  `a3eecadcf847ac70fc28c58dad642f2d761015e0`, but the first ENG-006 start failed
+  because authority validation mixed ENG-006 lifecycle identity with ENG-007
+  protected-check evidence. No successor is active.
+- Review gate: all nine internal tracks passed exact implementation head
+  `c59bff278945c3d8e63a6ea776a6bf6206df8af8`; awaiting external checks and the
   user-owned merge decision.

.agent-loop/initiatives/WS-ENG-007-concurrent-pr-review-reconciliation/chunks/WS-ENG-007-00R4-cross-initiative-authority-projection.md

@@ -0,0 +1,95 @@
+# Chunk Contract: WS-ENG-007-00R4 — Cross-Initiative Authority Projection Repair
+
+## Parent initiative
+
+`WS-ENG-007` — Concurrent PR Review Reconciliation
+
+## Goal
+
+Restore signed starts for an idle initiative when the latest global merge belongs
+to another initiative, without weakening merge-bound evidence validation.
+
+## Why this chunk exists
+
+After PR #190 reconciled trusted `main`, the signed start of `WS-ENG-006-01`
+failed closed. Authority-event validation combined ENG-006's earlier signed
+lifecycle with ENG-007's latest protected-check evidence, then correctly rejected
+that synthetic record because the evidence belonged to a different PR.
+
+## Risk class
+
+L1 / P0 authority-event reliability repair.
+
+## Start phase
+
+Recovery implementation. The signed start mechanism fails before it can record
+this repair, so this contract documents the exact bounded recovery exception.
+
+## Allowed files
+
+```text
+scripts/update_post_merge_memory.py
+scripts/check_loop_memory_state.py
+scripts/test_update_post_merge_memory.py
+.agent-loop/initiatives/WS-ENG-007-concurrent-pr-review-reconciliation/**
+.agent-loop/merge-intents/WS-ENG-007-00R4.json
+```
+
+## Not allowed
+
+```text
+workflow, permission, dispatcher, branch-protection, or cancel-authority changes
+protected-check, internal-review, CI, coverage, or human-merge weakening
+application, API, database, auth, payment, or product changes
+automatic successor starts or unsigned implementation authority
+```
+
+## Acceptance criteria
+
+- [ ] Authority events retain the latest global merge record as their canonical
+      base and retain its exact protected-check evidence unchanged.
+- [ ] The selected initiative projection contains only its signed lifecycle
+      identity and active/gate transition; it does not borrow evidence from the
+      latest global PR.
+- [ ] Ledger transition validation still binds the authority source and completed
+      chunk byte-for-byte to the selected initiative's latest signed record.
+- [ ] Malformed or forged authority sources, initiative identities, chunk
+      identities, active state, and gates still fail closed.
+- [ ] A regression fixture proves an older idle initiative can start after a
+      different initiative's merge-bound evidence becomes the global state.
+- [ ] The independent checker applies the same separation and validates the
+      complete generated state and ledger.
+- [ ] Exactly one schema-v2 merge intent names `WS-ENG-007-01` as the stopped
+      explicit successor; neither ENG-006 nor ENG-007 starts automatically.
+
+## Verification commands
+
+```bash
+PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 python3 -m pytest -q scripts/test_update_post_merge_memory.py scripts/test_check_loop_memory_state.py
+python3 scripts/check_markdown_links.py
+python3 scripts/check_stale_workstream_wording.py
+python3 scripts/update_post_merge_memory.py validate-merge-intent --repository-root . --base-ref origin/main
+git diff --check origin/main...HEAD
+```
+
+## Required reviewers
+
+- senior engineering
+- QA/test
+- security/auth
+- product/ops
+- architecture
+- CI integrity
+- docs
+- reuse/dedup
+- test delta
+
+## Human review focus
+
+Does the repair separate global merge evidence from initiative-local lifecycle
+state while keeping both independently and immutably bound?
+
+## Stop conditions
+
+Stop if the repair requires weakening protected evidence, accepting an unsigned
+basis, changing start authority, or automatically starting any successor.

.agent-loop/initiatives/WS-ENG-007-concurrent-pr-review-reconciliation/reviews/WS-ENG-007-00R4-external-review-response.md

@@ -0,0 +1,35 @@
+# External Review Response: WS-ENG-007-00R4
+
+## Comments addressed
+
+- GitHub Agent Gates failed because independent checker branch coverage was
+  89.13 percent, below the protected 90 percent floor.
+- Added seven focused fail-closed mutations for invalid PR identity, canonical
+  PR URL, bounded title, malformed and timezone-naive timestamps, and authority
+  schema shape.
+- The exact failing GitHub command now passes 296 tests with 90.40 percent
+  checker branch coverage.
+- CodeRabbit identified a stale chunk-map dependency that still named `00R3`.
+  The `WS-ENG-007-01` row now names `00R4` and successful reconciliation as its
+  prerequisites.
+
+## Comments deferred
+
+None.
+
+## Human decisions needed
+
+None. This response changes tests and process documentation only; it does not
+alter CI, thresholds, workflow behavior, permissions, or authority semantics.
+
+## Commands rerun
+
+```bash
+PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 python3 -m pytest -p pytest_cov.plugin -q --cov=scripts.check_loop_memory_state --cov-branch --cov-report=term-missing --cov-fail-under=90 scripts/test_agent_gates.py scripts/test_update_post_merge_memory.py scripts/test_check_loop_memory_state.py
+git diff --check
+```
+
+## Remaining risks
+
+The updated exact PR head must complete GitHub Agent Gates, Backend, and
+CodeRabbit before the PR is ready for human merge review.

.agent-loop/initiatives/WS-ENG-007-concurrent-pr-review-reconciliation/reviews/WS-ENG-007-00R4-internal-review-evidence.md

@@ -0,0 +1,77 @@
+# Internal Review Evidence: WS-ENG-007-00R4
+
+## Chunk
+
+`WS-ENG-007-00R4` — Cross-Initiative Authority Projection Repair
+
+open sub-agent sessions: none
+
+valid findings addressed: yes
+
+## Reviewed Revision
+
+Reviewed code SHA: c59bff278945c3d8e63a6ea776a6bf6206df8af8
+
+Reviewed at: 2026-07-23T10:50:17Z
+
+The `/root/eng006_*` sessions were inherited from the ENG-006 reviewer pool and
+explicitly reassigned to this WS-ENG-007-00R4 review. Every track is rebound to
+the exact reviewed implementation SHA above; session names grant no
+cross-initiative authority.
+
+Reviewer run IDs: senior-engineering=/root/eng006_senior_arch_docs; QA/test=/root/eng006_qa_ci_tests; security/auth=/root/eng006_security_ops_reuse; product/ops=/root/eng006_security_ops_reuse; architecture=/root/eng006_senior_arch_docs; docs=/root/eng006_senior_arch_docs; CI-integrity=/root/eng006_qa_ci_tests; reuse/dedup=/root/eng006_security_ops_reuse; test-delta=/root/eng006_qa_ci_tests
+
+## Reviewer Results
+
+| Reviewer | Result | Blocking findings | Notes |
+|---|---:|---|---|
+| senior engineering | PASS AFTER FIXES | None | Authority lifecycle and global evidence remain separate and ledger-bound. |
+| QA/test | PASS AFTER FIXES | None | Exact cross-initiative regression and malformed-basis cases pass. |
+| security/auth | PASS | None | Start authority, permissions, and signed-basis binding are unchanged. |
+| product/ops | PASS | None | Per-initiative concurrency remains isolated; no product lifecycle change. |
+| architecture | PASS AFTER FIXES | None | Removed invalid synthetic record composition without adding a parallel path. |
+| CI integrity | PASS AFTER FIXES | None | 296 tests pass; updater and checker retain literal protected coverage floors. |
+| docs | PASS | None | Contract, chunk map, and status describe the bounded recovery. |
+| reuse/dedup | PASS | None | Existing transition validators remain the single ledger-binding path. |
+| test delta | PASS AFTER FIXES | None | Additive adversarial tests; no skips, removals, or weakened assertions. |
+
+## Valid Findings Addressed
+
+- Prevented the independent checker from crashing on non-object authority
+  metadata and restored explicit source scalar, timestamp, PR, metadata, and
+  intent-path validation.
+- Added an independent checker regression for an older idle initiative starting
+  after a different initiative's protected merge became global state.
+- Added six updater mutation cases for malformed authority source and completed
+  metadata.
+- Raised exact updater branch coverage from 89.96 percent to 90.18 percent
+  without rounding, threshold changes, exclusions, or test weakening.
+- Responded to the external Agent Gates failure by raising independent checker
+  branch coverage from 89.13 percent to 90.40 percent with seven adversarial
+  cases and no threshold or workflow change.
+
+## Commands Run
+
+```bash
+PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 python3 -m pytest -q scripts/test_update_post_merge_memory.py scripts/test_check_loop_memory_state.py scripts/test_agent_gates.py
+PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 python3 -m pytest -p pytest_cov -q scripts/test_update_post_merge_memory.py scripts/test_check_loop_memory_state.py scripts/test_agent_gates.py --cov=scripts.update_post_merge_memory --cov-branch --cov-report=term --cov-fail-under=90 --cov-precision=2
+python3 scripts/check_markdown_links.py
+python3 scripts/check_stale_workstream_wording.py
+python3 scripts/update_post_merge_memory.py validate-merge-intent --repository-root . --base-ref origin/main
+git diff --check
+```
+
+The exact updater gate passed 289 tests at 90.18 percent branch coverage. The
+exact independent-checker gate passed 296 tests at 90.40 percent branch
+coverage. Both retain the literal 90.00 percent blocking floor.
+
+## Remaining Risks
+
+The current failed start run remains unsigned and inert. After this recovery
+merges and generated memory succeeds, ENG-006 still requires a fresh explicit
+start against exact current `main`; nothing starts automatically.
+
+## Stop Condition
+
+No successor is active. `WS-ENG-007-01` and `WS-ENG-006-01` each retain their
+own explicit signed start gate.

.agent-loop/initiatives/WS-ENG-007-concurrent-pr-review-reconciliation/reviews/WS-ENG-007-00R4-pr-trust-bundle.md

@@ -0,0 +1,45 @@
+# PR Trust Bundle: WS-ENG-007-00R4
+
+## Intent
+
+Allow a signed start for an idle initiative after another initiative becomes the
+latest global merge, without attaching one PR's protected evidence to another
+initiative's lifecycle.
+
+## Scope
+
+- Keep the latest global merge and its protected evidence unchanged.
+- Validate the selected initiative's authority projection independently.
+- Bind that projection byte-for-byte to its prior signed ledger basis.
+- Add updater and independent-checker regressions for the exact failure.
+
+## Reviewed Revision
+
+`c59bff278945c3d8e63a6ea776a6bf6206df8af8`
+
+## CI Integrity
+
+- No workflow, permission, required check, coverage floor, test, reviewer, signed
+  start, or human merge checkpoint was removed or weakened.
+- The exact updater gate passes 289 tests at 90.18 percent branch coverage.
+- The exact independent-checker gate passes 296 tests at 90.40 percent branch
+  coverage.
+
+## Reviewer Result
+
+All nine required internal tracks passed after fixes. No Critical or High
+finding remains open.
+
+## Human Review Focus
+
+- Confirm authority projection validation no longer constructs a synthetic merge
+  record with foreign protected evidence.
+- Confirm source/completed identity remains bound to the prior signed initiative
+  record by both transition validators.
+- Confirm malformed authority metadata fails closed in updater and checker.
+- Confirm no successor starts automatically.
+
+## Remaining Risk
+
+A new `main` merge before this recovery lands requires the branch and evidence to
+be refreshed normally; the signed event still requires exact current `main`.

.agent-loop/merge-intents/WS-ENG-007-00R4.json

@@ -0,0 +1,9 @@
+{
+  "chunk_id": "WS-ENG-007-00R4",
+  "chunk_title": "Cross-Initiative Authority Projection Repair",
+  "initiative_id": "WS-ENG-007",
+  "next_chunk_id": "WS-ENG-007-01",
+  "next_chunk_title": "Reviewed Patch and Base-Delta Reconciliation",
+  "next_requires_explicit_start": true,
+  "schema_version": 2
+}

scripts/check_loop_memory_state.py

@@ -465,31 +465,64 @@ def _record_failures(
         authority = base.pop("authority_state", None)
         metadata = base.get("completed_chunk", {})
         source = base.get("source", {})
+        repository = base.get("repository")
         base["updated_at"] = source.get("merged_at")
         failures = _record_failures(base, label, repository_root)
         if not isinstance(authority, dict) or set(authority) != {
             "source", "completed_chunk", "active", "gate"
         }:
             failures.append(f"{label}: invalid authority lifecycle state")
             return failures
+        authority_source = authority["source"]
+        if not isinstance(authority_source, dict) or set(authority_source) != {
+            "main_sha", "first_parent_sha", "pr_number", "pr_url", "pr_title",
+            "head_sha", "head_ref", "merged_at", "merged_by", "intent_path",
+            "intent_blob_sha",
+        }:
+            failures.append(f"{label}: invalid authority lifecycle source")
+            return failures
+        for field in ("main_sha", "first_parent_sha", "head_sha", "intent_blob_sha"):
+            value = authority_source.get(field)
+            if not isinstance(value, str) or not SHA_PATTERN.fullmatch(value):
+                failures.append(f"{label}: invalid authority source {field}")
+        authority_pr = authority_source.get("pr_number")
+        if (
+            not isinstance(authority_pr, int)
+            or isinstance(authority_pr, bool)
+            or authority_pr <= 0
+        ):
+            failures.append(f"{label}: invalid authority source pr_number")
+        elif authority_source.get("pr_url") != (
+            f"https://github.com/{repository}/pull/{authority_pr}"
+        ):
+            failures.append(f"{label}: invalid authority source pr_url")
+        for field, maximum in (
+            ("pr_title", 240), ("head_ref", 240), ("merged_by", 160)
+        ):
+            if not _is_bounded_single_line(authority_source.get(field), maximum):
+                failures.append(f"{label}: invalid authority source {field}")
+        try:
+            authority_time = datetime.fromisoformat(
+                authority_source["merged_at"].replace("Z", "+00:00")
+            )
+        except (AttributeError, ValueError):
+            failures.append(f"{label}: invalid authority source merged_at")
+        else:
+            if authority_time.tzinfo is None:
+                failures.append(f"{label}: authority source merged_at has no timezone")
         metadata = authority["completed_chunk"]
-        lifecycle = json.loads(json.dumps(base))
-        lifecycle.update(authority)
-        lifecycle_source = lifecycle.get("source", {})
-        lifecycle["updated_at"] = lifecycle_source.get("merged_at")
-        lifecycle_metadata = lifecycle.get("completed_chunk", {})
-        lifecycle["active"] = {"planning_chunk": None, "implementation_chunk": None}
-        lifecycle["gate"] = {
-            "status": "stopped_after_merge",
-            "next_chunk_id": lifecycle_metadata.get("next_chunk_id"),
-            "next_chunk_title": lifecycle_metadata.get("next_chunk_title"),
-            "next_requires_explicit_start": lifecycle_metadata.get(
-                "next_requires_explicit_start"
-            ),
-        }
-        failures.extend(
-            _record_failures(lifecycle, f"{label} authority basis", repository_root)
-        )
+        metadata_failures = _metadata_failures(metadata, f"{label} authority basis")
+        failures.extend(metadata_failures)
+        if not isinstance(metadata, dict):
+            return failures
+        if authority_source.get("intent_path") != (
+            f".agent-loop/merge-intents/{metadata.get('chunk_id')}.json"
+        ):
+            failures.append(f"{label}: authority intent path does not match completed chunk")
+        # The authority projection is bound independently to its exact prior
+        # initiative record by _authority_transition_failures.  Rebuilding a
+        # synthetic merge record here would incorrectly attach the latest
+        # global PR's protected-check evidence to that older initiative basis.
         historical_event_keys = {
             "type", "event_id", "run_id", "created_at", "dispatcher",
             "approvers", "reason", "main_sha", "prior_state_tip",