fix(pr-review): require repository reuse evidence before approval

examples/pr-review-command-smoke.py

@@ -267,7 +267,7 @@ def fake_run_gh_json(args: list[str], *, cwd: Path | None = None) -> object:
         if req.get("evidence_id") == "scope_fit"
     ]
     assert scope_fit_reqs, "scope_fit evidence requirement missing from contract"
-    assert scope_fit_reqs[0]["required_when"] == "code_change", scope_fit_reqs
+    assert scope_fit_reqs[0]["required_when"] == "behavior_bearing_change", scope_fit_reqs
     proportionality_reqs = [
         req
         for req in contract.get("evidence_requirements", [])
@@ -288,7 +288,7 @@ def fake_run_gh_json(args: list[str], *, cwd: Path | None = None) -> object:
         if req.get("evidence_id") == "default_off_isolation"
     ]
     assert isolation_reqs, "default_off_isolation evidence requirement missing"
-    assert isolation_reqs[0]["required_when"] == "code_change"
+    assert isolation_reqs[0]["required_when"] == "behavior_bearing_change"
     assert "paired_counterfactual_validation" in isolation_reqs[0]["fields"]
     authority_reqs = [
         req
@@ -321,6 +321,9 @@ def fake_run_gh_json(args: list[str], *, cwd: Path | None = None) -> object:
     assert code_pr["review_plan"]["applicability"]["scope_fit_required"] is True, (
         code_pr["review_plan"]
     )
+    reuse = code_pr["review_plan"]["result_template"]["evidence"]["repository_reuse"]
+    assert reuse == {"status": "unverified"}, reuse
+    assert code_pr["review_plan"]["applicability"]["repository_reuse_required"] is True
     assert (
         code_pr["review_plan"]["applicability"]["change_proportionality_required"]
         is True
@@ -343,6 +346,7 @@ def fake_run_gh_json(args: list[str], *, cwd: Path | None = None) -> object:
     )
     assert docs_pr is not None, "fixture must include a docs-only PR"
     assert docs_pr["review_plan"]["applicability"]["scope_fit_required"] is False
+    assert "repository_reuse" not in docs_pr["review_plan"]["required_evidence_ids"]
     assert (
         docs_pr["review_plan"]["applicability"]["change_proportionality_required"]
         is False
@@ -686,6 +690,7 @@ def fake_run_gh_json(args: list[str], *, cwd: Path | None = None) -> object:
     assert set(requirements) == {
         "problem_context",
         "architecture_flow",
+        "repository_reuse",
         "changed_line_classification",
         "scope_fit",
         "symbol_map",
@@ -752,6 +757,7 @@ def fake_run_gh_json(args: list[str], *, cwd: Path | None = None) -> object:
     assert execution["completion_gate"]["metadata_only_verdict_allowed"] is False
     assert execution["completion_gate"]["stale_head_verdict_allowed"] is False
     assert execution["completion_gate"]["blocking_evidence_verdicts"] == {
+        "repository_reuse": ["unjustified_duplication", "not_yet_proven"],
         "change_proportionality": ["disproportionate", "not_yet_proven"],
         "default_off_isolation": ["not_isolated", "not_yet_proven"],
         "authority_semantics": ["misleading", "not_yet_proven"],

loopx/capabilities/pr_review_queue/README.md

@@ -178,6 +178,14 @@ typed evidence groups before a verdict:
 
 - problem context and active caller;
 - architecture and ownership flow;
+- repository reuse for behavior-bearing changes: search base and exact-head
+  code, including unchanged siblings, by caller outcome/resource rather than
+  only new filenames. Record revisions, queries, paths and candidate callers;
+  compare scope/filters, ordering/paging, authority/sanitization and state/retry
+  ownership. Prefer the nearest existing owner; justify independent boundaries
+  with evidence, not green CI or conflict-free coexistence. For alternative
+  views of one resource, validate consumer switching and concurrent updates
+  where relevant. Repeat the comparison after base integration or head changes;
 - exact changed-line classification across production, tests/fixtures, docs,
   generated output, and mechanical moves;
 - a 2-5 item exact-head symbol map for code-changing PRs, including caller,
@@ -221,6 +229,18 @@ hints, and green CI cannot upgrade evidence to `verified`. A stale-head verdict
 is prohibited. Missing evidence remains `unverified` with a reason instead of
 being replaced by confident prose.
 
+`repository_reuse` starts unverified in every applicable plan. Its conclusions
+are `reused`, `separation_justified`, `no_existing_candidate`,
+`unjustified_duplication`, or `not_yet_proven`. A negative search must name its
+scope and limitations; an empty candidate list is not proof of absence.
+Unjustified duplication or missing evidence requires a request-changes
+conclusion. Similar-looking code with distinct invariants or compatibility
+needs may legitimately remain separate. Ordinary docs retain their existing
+review path; smoke-only changes retain `durable_smoke_value` coverage review.
+This is a reviewer-executed contract projected by the packet, not an automatic
+repository search or a semantic validator of published prose. Tests establish
+packet applicability and verdict policy, not guaranteed model compliance.
+
 When `--state all` is used, the command must preserve both lifecycle groups.
 The `--limit` value is applied per group so a busy open queue cannot consume the
 whole packet and make `review_groups.merged` empty while merged PRs exist in the

loopx/capabilities/pr_review_queue/catalog_entry.py

@@ -110,7 +110,7 @@
     ],
     "docs": ["loopx/capabilities/pr_review_queue/README.md"],
     "boundaries": [
-        "The shared execution contract owns review depth, evidence completeness, exact-head freshness, symbol-map, walkthrough, validation, failure, code-volume, change-proportionality, default-off isolation, and authority-semantics requirements; host skills only route and publish it.",
+        "The shared execution contract owns review depth, evidence completeness, repository-reuse comparison, exact-head freshness, symbol-map, walkthrough, validation, failure, code-volume, change-proportionality, default-off isolation, and authority-semantics requirements; host skills only route and publish it.",
         "A queue is observed only when result_completeness.complete=true; partial or failed reads are not_observed and never count as unchanged.",
         "Fingerprints cover exact head, formal conclusion, next action, check state, draft state, and mergeability for every open PR.",
         "Current-head review_ready_at, not updatedAt, owns age-fair ordering; one new head after REQUEST_CHANGES may use a bounded fast-feedback slot.",

loopx/capabilities/pr_review_queue/review_contract.py

@@ -76,7 +76,7 @@ def build_review_template(item: Mapping[str, Any]) -> dict[str, Any]:
             _section(
                 "改动思路",
                 "300-500字",
-                "Use `architecture_flow` and `walkthroughs`: entry point, authoritative state, decision boundary, positive path, alternative, and ownership trade-off.",
+                "Use `architecture_flow`, `repository_reuse`, and `walkthroughs`: entry point, authoritative state, decision boundary, positive path, existing implementation comparison, and ownership trade-off.",
             ),
             _section(
                 "具体改动",
@@ -137,6 +137,48 @@ def build_review_execution_contract() -> dict[str, Any]:
                     "failure_or_retry_owner",
                 ],
             },
+            {
+                "evidence_id": "repository_reuse",
+                "required_when": "behavior_bearing_change",
+                "verdict_values": [
+                    "reused", "separation_justified", "no_existing_candidate",
+                    "unjustified_duplication", "not_yet_proven",
+                ],
+                "fields": [
+                    "searched_revisions", "queries_and_paths", "existing_candidates",
+                    "semantic_comparison", "reuse_or_separation_reason",
+                    "validation_evidence", "verdict",
+                ],
+                "comparison_dimensions": [
+                    "resource_and_caller", "data_scope_and_filters",
+                    "ordering_and_pagination", "authority_and_sanitization",
+                    "state_retry_and_failure_owner",
+                ],
+                "rule": (
+                    "Before judging the proposed architecture, search the base and "
+                    "exact-head repository by caller outcome, resource, routes and "
+                    "symbols, not only new filenames. Record immutable searched "
+                    "revisions, queries, paths and candidate definitions/callers, "
+                    "including unchanged siblings outside the diff. Compare the "
+                    "applicable dimensions for each candidate and explain why the "
+                    "nearest existing owner can be reused or extended, or why an "
+                    "independent boundary is necessary. Same-resource views must "
+                    "retain the same semantics unless a deliberate difference is "
+                    "disclosed and validated; exercise switching consumers and "
+                    "concurrent updates when state or paging is involved. Green CI, "
+                    "shared low-level readers, different names, and conflict-free "
+                    "coexistence do not justify a second contract or state owner. "
+                    "A no_existing_candidate verdict requires a bounded negative "
+                    "search with its scope and limitations, not an empty candidate "
+                    "list. Similar code alone is not duplication: preserve distinct "
+                    "invariants or compatibility boundaries when evidence justifies "
+                    "separation. After base integration or head changes, repeat the "
+                    "comparison rather than inheriting an earlier approval. Missing "
+                    "or unverified evidence is not_yet_proven and cannot approve. "
+                    "This is a reviewer-executed evidence requirement, not an "
+                    "automatic similarity detector or proof that a search ran."
+                ),
+            },
             {
                 "evidence_id": "changed_line_classification",
                 "required_when": "always",
@@ -507,6 +549,7 @@ def build_review_execution_contract() -> dict[str, Any]:
             "exact_head_recheck_required": True,
             "stale_head_verdict_allowed": False,
             "blocking_evidence_verdicts": {
+                "repository_reuse": ["unjustified_duplication", "not_yet_proven"],
                 "change_proportionality": [
                     "disproportionate",
                     "not_yet_proven",
@@ -524,6 +567,11 @@ def build_review_execution_contract() -> dict[str, Any]:
         },
         "verdict_policy": {
             "open_pr_blocking_finding": "REQUEST_CHANGES",
+            "open_pr_unresolved_reuse": (
+                "REQUEST_CHANGES when repository_reuse is unjustified_duplication "
+                "or not_yet_proven, including missing/unverified search evidence; "
+                "request reuse, an evidence-backed separation, or further investigation"
+            ),
             "open_pr_unresolved_proportionality": (
                 "REQUEST_CHANGES when change_proportionality is disproportionate "
                 "or not_yet_proven; correctness, green CI, and resolved earlier "
@@ -580,6 +628,7 @@ def build_review_plan(item: Mapping[str, Any]) -> dict[str, Any]:
         required_evidence.append("authority_semantics")
         required_evidence.append("typed_state_rule")
     if behavior_bearing_change:
+        required_evidence.insert(3, "repository_reuse")
         if "scope_fit" not in required_evidence:
             required_evidence.append("scope_fit")
         required_evidence.append("default_off_isolation")
@@ -609,6 +658,7 @@ def build_review_plan(item: Mapping[str, Any]) -> dict[str, Any]:
             "docs_only": docs_only,
             "symbol_map_required": code_change,
             "scope_fit_required": behavior_bearing_change,
+            "repository_reuse_required": behavior_bearing_change,
             "change_proportionality_required": code_change,
             "default_off_isolation_required": behavior_bearing_change,
             "authority_semantics_required": code_change,

skills/loopx-pr-review/SKILL.md

@@ -125,20 +125,11 @@ artifacts back.
 
 ## Full PR Interpretation Depth
 
-A complete review is a whole-PR interpretation, not a checklist or findings
-summary. For each selected PR:
-
-1. Read every changed file and map each file to its responsibility, inputs,
-   outputs, and key symbols.
-2. Pick 2-5 behavior-bearing symbols and explain before/after behavior,
-   critical branches, callers/callees, side effects, and failure paths.
-3. Walk one positive path from user/host action to observable result.
-4. Walk one negative path (invalid input, permission, timeout, corrupt state,
-   private boundary, or rollback) and show where it fails closed.
-5. Cover all changed surfaces in the five sections: motivation, approach,
-   concrete changes, main risk, overall judgment.
-6. List validation per surface and name anything not independently verified.
-7. State overall judgment for the entire PR, not only for the top finding.
+Use the packet's `repository_reuse`, `symbol_map`, `walkthroughs`, and
+`validation_matrix` evidence across the whole PR, including unchanged callers
+and sibling implementations. Walk one positive path to the observable result.
+Walk one negative path and explain its failure owner. Render the verified
+evidence in the five sections; do not maintain a second checklist here.
 
 A review that only repeats the PR body, only discusses one blocker, or omits
 whole files/modules is incomplete and must be reworked.

tests/capabilities/test_pr_review_contract.py

@@ -1,5 +1,7 @@
 from __future__ import annotations
 
+import pytest
+
 from loopx.capabilities.pr_review_queue import (
     build_agent_response_contract,
     build_review_plan,
@@ -40,6 +42,7 @@ def test_execution_contract_owns_deep_review_requirements() -> None:
     assert set(requirements) == {
         "problem_context",
         "architecture_flow",
+        "repository_reuse",
         "changed_line_classification",
         "scope_fit",
         "symbol_map",
@@ -117,6 +120,7 @@ def test_execution_contract_owns_deep_review_requirements() -> None:
     assert contract["completion_gate"]["metadata_only_verdict_allowed"] is False
     assert contract["completion_gate"]["stale_head_verdict_allowed"] is False
     assert contract["completion_gate"]["blocking_evidence_verdicts"] == {
+        "repository_reuse": ["unjustified_duplication", "not_yet_proven"],
         "change_proportionality": ["disproportionate", "not_yet_proven"],
         "default_off_isolation": ["not_isolated", "not_yet_proven"],
         "authority_semantics": ["misleading", "not_yet_proven"],
@@ -265,3 +269,75 @@ def test_test_only_plan_skips_runtime_lenses() -> None:
     assert plan["applicability"]["guidance_vs_obligation_required"] is False
     assert "typed_state_rule" not in plan["required_evidence_ids"]
     assert "domain_neutrality" not in plan["required_evidence_ids"]
+
+
+@pytest.mark.parametrize(
+    "area",
+    [
+        "product_runtime",
+        "app_or_ui_surface",
+        "ci_or_release",
+        "build_or_config",
+        "agent_instruction_surface",
+        "public_entry_or_policy",
+    ],
+)
+def test_behavior_review_requires_repository_reuse_even_with_green_checks(
+    area: str,
+) -> None:
+    item = _item(areas={area: 1, "test_or_example": 1})
+    # A narrow changed-file list and passing CI cannot establish that an
+    # unchanged sibling already implements the same caller outcome.
+    item["key_files"] = [{"path": "src/history_list.py", "additions": 80}]
+    item["checks"] = {"counts": {"success": 4, "failure": 0}}
+    plan = build_review_plan(item)
+    assert plan["applicability"]["repository_reuse_required"] is True
+    assert "repository_reuse" in plan["required_evidence_ids"]
+    assert plan["result_template"]["evidence"]["repository_reuse"] == {
+        "status": "unverified"
+    }
+
+
+@pytest.mark.parametrize("area", ["public_docs", "test_or_example"])
+def test_non_behavior_review_keeps_existing_coverage_policy(area: str) -> None:
+    plan = build_review_plan(_item(areas={area: 1}))
+    assert plan["applicability"]["repository_reuse_required"] is False
+    assert "repository_reuse" not in plan["required_evidence_ids"]
+
+
+def test_reuse_evidence_compares_semantics_beyond_the_diff() -> None:
+    contract = build_agent_response_contract()["review_execution_contract"]
+    reuse = next(
+        row
+        for row in contract["evidence_requirements"]
+        if row["evidence_id"] == "repository_reuse"
+    )
+    assert reuse["required_when"] == "behavior_bearing_change"
+    assert reuse["verdict_values"] == [
+        "reused",
+        "separation_justified",
+        "no_existing_candidate",
+        "unjustified_duplication",
+        "not_yet_proven",
+    ]
+    assert {
+        "searched_revisions",
+        "queries_and_paths",
+        "existing_candidates",
+        "semantic_comparison",
+        "reuse_or_separation_reason",
+        "validation_evidence",
+        "verdict",
+    } <= set(reuse["fields"])
+    assert {
+        "resource_and_caller",
+        "data_scope_and_filters",
+        "ordering_and_pagination",
+        "authority_and_sanitization",
+        "state_retry_and_failure_owner",
+    } <= set(reuse["comparison_dimensions"])
+    assert "unchanged" in reuse["rule"]
+    assert "negative search" in reuse["rule"]
+    assert "coexistence" in reuse["rule"]
+    assert "not an automatic similarity detector" in reuse["rule"]
+    assert "repository_reuse" in contract["verdict_policy"]["open_pr_unresolved_reuse"]