Lane prompt Rule 2: status, license and popularity are not evidence of fit (D1 merit rule)

manifests/evidence.json

@@ -38331,6 +38331,11 @@
       "sha256": "0b5d0b94d14aad9db95a31f2f7bd015515fb26f2abe74a60254dbdd6244a8808",
       "bytes": 146176
     },
+    {
+      "path": "tests/test_lane_packets.py",
+      "sha256": "940485c85045360783b344debc7bc92caf74ea203c739108a3b0a2800174011b",
+      "bytes": 98784
+    },
     {
       "path": "tests/test_layer_verdicts.py",
       "sha256": "ee85d714bdd00edcc217e20441d0e59a2b3382c0533496428558d75b764c9267",
@@ -38636,6 +38641,11 @@
       "sha256": "a4236a158789418581150b454b996df3ae2d17826e89728e56f1cffbfa019b60",
       "bytes": 29398
     },
+    {
+      "path": "tests/test_verdict_lane_vendoring.py",
+      "sha256": "02465e7ca09a3d4ee72e3855084cc5b54ee223b209bb6169cb9deea1c5d9fdab",
+      "bytes": 12669
+    },
     {
       "path": "tests/test_worker_recovery.py",
       "sha256": "9360d522838325f232d6f1d7a02dcefa36a9865a0fa02c22bec42db606a8dc0a",
@@ -38851,10 +38861,15 @@
       "sha256": "de2377990fa015fe8666b8ebb9176968c5600a3f3f2c85c2a9f927c0af7d1cae",
       "bytes": 8214
     },
+    {
+      "path": "tools/sota-convergence/lane-prompt.md",
+      "sha256": "d3ef2cc0c06d33d6a4a058f5e74a1cf5be8a3760cc2b0b5572598f0e0e4288e4",
+      "bytes": 2220
+    },
     {
       "path": "tools/sota-convergence/lane-provenance.json",
-      "sha256": "ca747c9808c723822c5ef970967ff885d33b2834f37bc5ff5ff07301ec7f2176",
-      "bytes": 8231
+      "sha256": "8281bcc79f76dcaf7427b76429613cd45597d990e747f49a917d3dae9c93d261",
+      "bytes": 9569
     },
     {
       "path": "tools/sota-convergence/reconciliations-20260922.json",

tests/test_lane_packets.py

@@ -15,6 +15,7 @@
 import importlib.util
 import io
 import json
+import re
 import shutil
 import tempfile
 import unittest
@@ -47,6 +48,12 @@ def load_module(name, filename):
 
 # Seeded order of the manifest-mode fixture (seed 20260922, layer-b); newcomers have no component id.
 EXPECTED_SEEDED_COMPONENT_ORDER = [None, None, 'data-two', 'data-one', 'data-three']
+# Rule 2's merit sentence (2026-09-26 user directives, verdict-wave runbook D1): a winner earns it on what its
+# evidence shows was run, never on its status, license or popularity. Generic by design: the shared prompt names no
+# repository or product.
+MERIT_SENTENCE = ("Adoption, incumbency, installation, retained-control status, receipt count, packet position, "
+                  "license, stars and popularity are not evidence of fit; choose among adopted candidates on what "
+                  "their evidence shows was run.")
 
 
 class LanePacketsFixture(unittest.TestCase):
@@ -165,6 +172,8 @@ def test_rules_are_the_five_lane_prompt_rules(self):
         self.assertEqual(len(packet["rules"]), 5)
         self.assertEqual(packet["rules"], lane_packets.load_rules())
         self.assertIn("Judge from retained evidence.", packet["rules"][0])
+        # Every packet carries the merit rule, so the refuters and judges who read the packet see it too.
+        self.assertIn(MERIT_SENTENCE, packet["rules"][1])
 
     def test_group_is_present_for_us_equities_and_null_for_foundation(self):
         self.assertIsNone(self.packet("foundation", "layer-a")["group"])
@@ -191,6 +200,23 @@ def test_wrong_rule_count_raises(self):
                 lane_packets.load_rules(prompt)
 
 
+class LanePromptMeritRuleTests(unittest.TestCase):
+    """The shared prompt's Rules section (2026-09-26, verdict-wave runbook D1): Rule 2 carries the merit sentence,
+    and the file numbers exactly five rules, 1 to 5, counted here apart from load_rules (which only counts them), so
+    dropping the sentence, moving it out of Rule 2, or adding, removing or renumbering a rule fails."""
+
+    def test_rule_2_states_the_merit_sentence(self):
+        rules = lane_packets.load_rules()
+        self.assertTrue(rules[1].startswith("The winner set is 1-3 adopted candidates (adopted == true)"), rules[1])
+        self.assertIn(MERIT_SENTENCE, rules[1])
+
+    def test_the_rules_section_numbers_exactly_five_rules(self):
+        text = (TOOL_DIR / "lane-prompt.md").read_text(encoding="utf-8")
+        _, separator, section = text.partition("\nRules\n")
+        self.assertTrue(separator, "lane-prompt.md must keep its Rules section")
+        self.assertEqual(re.findall(r"(?m)^(\d+)\.\s", section), ["1", "2", "3", "4", "5"])
+
+
 class ShuffleTests(LanePacketsFixture):
     def test_shuffle_is_seeded_and_stable_across_repeated_builds(self):
         first = self.packet("foundation", "layer-a")

tests/test_verdict_lane_vendoring.py

@@ -91,6 +91,29 @@ def test_the_current_transcript_audit_code_is_registered_for_the_claude_lane(sel
                                 "register the current transcript_audit.py hash in tools/sota-convergence/"
                                 "lane-provenance.json's claude entries")
 
+    def test_the_current_claude_lane_key_is_registered_under_both_workflow_paths(self):
+        """2026-09-26 merit rule: claude_lane.py binds a Claude return's prompt_sha256 to this checkout's
+        lane-prompt.md, and record_verdicts.py and scripts/landscape.py refuse a return whose full key
+        (LANE_PROVENANCE_KEYS["claude"]) is not listed, so a lane-prompt.md edit needs Claude appends as well as the
+        Codex one test_the_current_codex_lane_code_is_registered checks."""
+        from scripts.landscape import LANE_PROVENANCE_KEYS, registered_provenance_entry
+        pin = json.loads((WORKFLOWS / "vendored-lanes.json").read_text(encoding="utf-8"))
+        item = next(entry for entry in pin["files"] if entry["path"] == "layer-verdict-lane.js")
+        vendored = f"examples/claude-native/workflows/{item['path']}"
+        current = {"workflow_sha256": sums()[item["path"]],
+                   "agent_sha256": sha256(ROOT / "examples" / "claude-native" / "agents" / "blind-lane-reviewer.md"),
+                   "prompt_sha256": sha256(TOOLS / "lane-prompt.md"),
+                   "transcript_audit_py_sha256": sha256(TOOLS / "transcript_audit.py")}
+        registry = {"claude": self.registry()["claude"]}
+        for workflow_path in (item["source_path"], vendored):
+            with self.subTest(workflow_path=workflow_path):
+                provenance = {"workflow_path": workflow_path, **current}
+                self.assertEqual(set(provenance), set(LANE_PROVENANCE_KEYS["claude"]))
+                entry = registered_provenance_entry("claude", provenance, registry)
+                self.assertIsNotNone(entry, "append the current Claude lane key, the lane-prompt.md hash included, to "
+                                            "tools/sota-convergence/lane-provenance.json under both workflow paths")
+                self.assertEqual(entry.get("vendored_path"), vendored)
+
     def test_the_lane_return_schema_admits_every_provenance_field(self):
         """Round 10, BR10-4: the provenance object sets additionalProperties false, so it must list every field a
         new-wave return carries."""
@@ -143,11 +166,12 @@ def test_registry_entries_are_well_formed_and_unique(self):
         registry = self.registry()
         # agent_sha256 (the blind-lane-reviewer role hash) is part of the Claude key from 2026-09-23; entries
         # registered before it simply lack it (the registry is append-only).
-        # An appended registration that changes only the audit code is a new key (round 11, RR11-2).
-        from scripts.landscape import ADJUDICATION_PROVENANCE_KEYS
-        for lane, fields in (("claude", ("workflow_path", "workflow_sha256", "agent_sha256",
-                                         "transcript_audit_py_sha256")),
-                             ("codex", ("codex_lane_py_sha256", "prompt_sha256")),
+        # An appended registration that changes only the audit code is a new key (round 11, RR11-2), and so is one
+        # that changes only lane-prompt.md (2026-09-26 merit rule): each lane's key is the one record time matches
+        # on, scripts/landscape.py's LANE_PROVENANCE_KEYS.
+        from scripts.landscape import ADJUDICATION_PROVENANCE_KEYS, LANE_PROVENANCE_KEYS
+        for lane, fields in (("claude", LANE_PROVENANCE_KEYS["claude"]),
+                             ("codex", LANE_PROVENANCE_KEYS["codex"]),
                              ("adjudication", ADJUDICATION_PROVENANCE_KEYS)):
             keys = [tuple(entry.get(field) for field in fields) for entry in registry[lane]]
             self.assertEqual(len(keys), len(set(keys)), lane)

tools/sota-convergence/lane-prompt.md

@@ -9,6 +9,8 @@ Rules
    Read only the packet and files under the repository root: no memory stores, code indexes, MCP tools, web
    search, git history, or other checkouts or work directories, which can carry the verdict you must reach.
 2. The winner set is 1-3 adopted candidates (adopted == true) that best satisfy the requirement on the evidence.
+   Adoption, incumbency, installation, retained-control status, receipt count, packet position, license, stars and
+   popularity are not evidence of fit; choose among adopted candidates on what their evidence shows was run.
    why_selected must cite at least one evidence path and state what was actually observed (native execution,
    measured comparison, local integration, synthetic fixture or source review), not what the project claims.
 3. Every adopted non-winner candidate appears in alternatives with a why_not_default that states the concrete

tools/sota-convergence/lane-provenance.json

@@ -88,6 +88,24 @@
    "workflow_path": "examples/claude-native/workflows/layer-verdict-lane.js",
    "workflow_sha256": "fd77b74945e9d2bd0b26b7822aabcde00e7e650c327da16d852fdbfd5984c0d6",
    "transcript_audit_py_sha256": "5ed7a1b58d2046a8b7c788026769d4a8a340e1ae92e2acf7ed06f03f12736b92"
+  },
+  {
+   "agent_sha256": "a87a9883618b320c8b97a442a43c471c0994fb70ea9bb64ce0b5f513ba31500c",
+   "prompt_sha256": "d3ef2cc0c06d33d6a4a058f5e74a1cf5be8a3760cc2b0b5572598f0e0e4288e4",
+   "registered_at": "2026-09-26",
+   "vendored_path": "examples/claude-native/workflows/layer-verdict-lane.js",
+   "workflow_path": ".claude/workflows/layer-verdict-lane.js",
+   "workflow_sha256": "fd77b74945e9d2bd0b26b7822aabcde00e7e650c327da16d852fdbfd5984c0d6",
+   "transcript_audit_py_sha256": "5ed7a1b58d2046a8b7c788026769d4a8a340e1ae92e2acf7ed06f03f12736b92"
+  },
+  {
+   "agent_sha256": "a87a9883618b320c8b97a442a43c471c0994fb70ea9bb64ce0b5f513ba31500c",
+   "prompt_sha256": "d3ef2cc0c06d33d6a4a058f5e74a1cf5be8a3760cc2b0b5572598f0e0e4288e4",
+   "registered_at": "2026-09-26",
+   "vendored_path": "examples/claude-native/workflows/layer-verdict-lane.js",
+   "workflow_path": "examples/claude-native/workflows/layer-verdict-lane.js",
+   "workflow_sha256": "fd77b74945e9d2bd0b26b7822aabcde00e7e650c327da16d852fdbfd5984c0d6",
+   "transcript_audit_py_sha256": "5ed7a1b58d2046a8b7c788026769d4a8a340e1ae92e2acf7ed06f03f12736b92"
   }
  ],
  "codex": [
@@ -130,6 +148,11 @@
    "codex_lane_py_sha256": "a4bd8f47138448ee80b3fbde6c4bdbf0a22185643efedcdf1573a8b3c596069e",
    "prompt_sha256": "9b8a8364f13ff3fcb9fc690198c4ca8849c29cc12b81e8526121a46114517f54",
    "registered_at": "2026-09-24"
+  },
+  {
+   "codex_lane_py_sha256": "a4bd8f47138448ee80b3fbde6c4bdbf0a22185643efedcdf1573a8b3c596069e",
+   "prompt_sha256": "d3ef2cc0c06d33d6a4a058f5e74a1cf5be8a3760cc2b0b5572598f0e0e4288e4",
+   "registered_at": "2026-09-26"
   }
  ],
  "schema_version": 1,