chore(bench): archive reference-only scripts

A1-A3 of the legacy inventory, moved with git mv to benchmarks/archive/:
the stage-2 verifier chain (verify_batched, score_batched, verify_sinks,
model_sinks, sink_candidates), the v2 tool hunter (hunt.py) and the
harnessx-removal equality proof. The plane's verify/agree/roles tasks
replaced the first six; the harnessx-removal spec is closed.

- The six independent scripts keep HERE = benchmarks/independent (population,
  evaluate.py, results) and add their own directory for archived siblings.
- harnessx_removal_equality's REPO root is one level deeper.
- tests/test_plane_scoring_parity.py loads score_batched from the archive
  (ran here, not skipped); test_harnessx_removal_equality points at the new path.
- benchmarks/archive/README.md: what each script was, which committed
  records cite it, the commit that last ran it. paths.md maps old -> new;
  committed records are not rewritten.

Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01EmM1j46qgKMh3zqD1QMwEo

benchmarks/archive/README.md

@@ -0,0 +1,26 @@
+# Archived benchmark scripts
+
+Reference-only scripts, moved here with `git mv` on 2026-10-02 (legacy cleanup) so their history
+follows them. Nothing in `src/` imports them and no current workflow runs them. They stay because
+committed records cite them and their numbers: to re-run a record, check out the commit the record
+names rather than running these copies against today's package.
+
+The old paths still appear inside committed records and in some docstrings. [paths.md](paths.md)
+maps each old path to the new one. The records themselves are not rewritten.
+
+| script | what it was | records that cite it | last recorded run |
+|---|---|---|---|
+| `verify_batched.py` | Stage-2 verifier, second design: triage per file, verify per file, report what two passes agree on. Replaced by the plane's `verify` + `agree` tasks. | `independent/results-v2-model-pipeline.json`, `independent/verifier-batched-check-2026-09-29.json`, `independent/plane-increment-2.json` (parity baseline) | `072a102` (named in `results-v2-model-pipeline.json`) |
+| `score_batched.py` | Exploratory scorer for the batched verifier: sites anchored to the candidate, matched by `evaluate.matches_v2`. | `independent/verifier-batched-check-2026-09-29.json`, `independent/plane-increment-2.json`; `tests/test_plane_scoring_parity.py` still loads it as the reference | added in `6acd354` (2026-09-29); the parity test replays it on hosts with the recorded scans |
+| `verify_sinks.py` | The first stage-2 verifier, one candidate at a time. `verify_batched.py` imports `ACCOUNT_ERRORS` and `_Recording` from it. | `independent/verifier-check-2026-09-29.json` | the known-answer check after `041d871` |
+| `model_sinks.py` | Stage-1 source-only model classification of sinks. Moved into the package as `plane/tasks/roles.py`. | `independent/selection-v2-model-sinks.json`, `measurements/2026-10-01-decision-engine-harvest/record.json`, both `2026-10-01-decision-engine-leak-audit` records | `05d5699` (named in `selection-v2-model-sinks.json`) |
+| `sink_candidates.py` | Diagnostic: vocabulary-declared sink candidates per case. Its declaration patterns live on in `plane/tasks/repo_facts.py`. | `independent/results-v2-model-pipeline.json` ("vocabulary sink-first: 2/17") | no commit recorded |
+| `hunt.py` | The v2 tool-hunter run over the population (protocol `hunter-protocol-v2.md`, including its `FAMILY_WORDS` table). Replaced by the plane's `verify` / `agree` tasks. | `independent/hunter-protocol-v2.md`, `independent/results-v2-hunter.json` (a label source in `learn/sources.toml`) | `6bfb05f`, the commit that introduced the protocol, which pins the runner |
+| `harnessx_removal_equality.py` | One-off equality proof for `harnessx-removal` (record and compare deterministic outputs). The spec is closed. `tests/test_harnessx_removal_equality.py` still checks that the comparison detects differences. | `measurements/2026-09-30-harnessx-removal-baseline/`, `measurements/2026-09-30-harnessx-removal-equality/` | baseline `ca5caf7`, candidate `b9afb66` |
+
+The six `independent` scripts still read the population, `evaluate.py` and the results from
+`benchmarks/independent/`. Their `HERE` points there, so outputs land where they always did.
+`evaluate.py` and `freeze.py` are current and were not moved.
+
+`verify_sinks.py --help` fails with an argparse `%` formatting error. That bug was already there
+before the move. The script still imports cleanly.

benchmarks/archive/harnessx_removal_equality.py

@@ -5,8 +5,8 @@
 non-zero on the first difference outside the exclusion list, naming the file and key.
 
 Usage:
-    python benchmarks/harnessx_removal_equality.py record <out-dir> [--tree-ish HEAD] [--workdir DIR]
+    python benchmarks/archive/harnessx_removal_equality.py record <out-dir> [--tree-ish HEAD] [--workdir DIR]
-    python benchmarks/harnessx_removal_equality.py compare <baseline-dir> <candidate-dir> \\
+    python benchmarks/archive/harnessx_removal_equality.py compare <baseline-dir> <candidate-dir> \\
         [--suite-map map.json] [--record comparison.json]
 
 The instrument checks come first and fail loudly: ``harnessx`` must not be importable, ``tree_sitter`` must be,
@@ -30,7 +30,7 @@ import xml.etree.ElementTree as ET
 from pathlib import Path
 from typing import Any
 
-REPO = Path(__file__).resolve().parents[1]
+REPO = Path(__file__).resolve().parents[2]  # archived under benchmarks/archive/
 
 # Design section 3 "Comparison": timestamps, durations and run identifiers are excluded; nothing else is.
 EXCLUDED_KEYS = frozenset({"runtime_seconds", "scan_id", "timestamp", "started", "finished", "created_at", "generated_at"})

benchmarks/archive/hunt.py

@@ -10,9 +10,9 @@ Model calls go to the resolved hunter client (DeepSeek when its key is set); the
 the recorded spend reaches `--budget-usd`.
 
 Usage:
-    python benchmarks/independent/hunt.py --population population-v2.toml [--only ID,...] [--pins a,b]
+    python benchmarks/archive/hunt.py --population population-v2.toml [--only ID,...] [--pins a,b]
-    python benchmarks/independent/hunt.py --population population-v2.toml --score
+    python benchmarks/archive/hunt.py --population population-v2.toml --score
-    python benchmarks/independent/hunt.py --pilot-root <checkout> --family injection   # a development case
+    python benchmarks/archive/hunt.py --pilot-root <checkout> --family injection   # a development case
 """
 
 from __future__ import annotations
@@ -25,7 +25,8 @@ import sys
 import time
 from pathlib import Path
 
-HERE = Path(__file__).resolve().parent
+# Archived 2026-10-02 from benchmarks/independent/: HERE stays that directory (population, evaluate.py, results).
+HERE = Path(__file__).resolve().parents[1] / "independent"
 sys.path.insert(0, str(HERE))
 
 import evaluate  # noqa: E402

benchmarks/archive/model_sinks.py

@@ -12,7 +12,7 @@ counted as UNCLASSIFIED, never as "no sinks". Hard spend ceiling `--budget-usd`;
 
 v2 is spent for tuning; this informs the selection design and qualifies nothing.
 
-Usage: python benchmarks/independent/model_sinks.py [--population population-v2.toml] [--only ID,...]
+Usage: python benchmarks/archive/model_sinks.py [--population population-v2.toml] [--only ID,...]
 """
 
 from __future__ import annotations
@@ -27,7 +27,8 @@ import threading
 import time
 from pathlib import Path
 
-HERE = Path(__file__).resolve().parent
+# Archived 2026-10-02 from benchmarks/independent/: HERE stays that directory (population, evaluate.py, results).
+HERE = Path(__file__).resolve().parents[1] / "independent"
 sys.path.insert(0, str(HERE))
 
 import evaluate  # noqa: E402

benchmarks/archive/paths.md

@@ -0,0 +1,24 @@
+# Old path to new path (2026-10-02)
+
+Committed records, specs and some docstrings name these scripts by their old path. Those files are
+not rewritten. Read the old path as the new one:
+
+| old path | new path |
+|---|---|
+| `benchmarks/independent/verify_batched.py` | `benchmarks/archive/verify_batched.py` |
+| `benchmarks/independent/score_batched.py` | `benchmarks/archive/score_batched.py` |
+| `benchmarks/independent/verify_sinks.py` | `benchmarks/archive/verify_sinks.py` |
+| `benchmarks/independent/model_sinks.py` | `benchmarks/archive/model_sinks.py` |
+| `benchmarks/independent/sink_candidates.py` | `benchmarks/archive/sink_candidates.py` |
+| `benchmarks/independent/hunt.py` | `benchmarks/archive/hunt.py` |
+| `benchmarks/harnessx_removal_equality.py` | `benchmarks/archive/harnessx_removal_equality.py` |
+
+Files that still use the old path:
+
+- Records: `independent/hunter-protocol-v2.md`, `independent/results-v2-model-pipeline.json`,
+  `independent/selection-v2-model-sinks.json`, `independent/verifier-batched-check-2026-09-29.json`,
+  `measurements/2026-09-30-harnessx-removal-equality/README`
+- Specs: `.kiro/specs/ai-service-plane/`, `.kiro/specs/harnessx-removal/design.md`,
+  `.kiro/specs/learned-decision-engine/`
+- Docstrings: `src/openultrasast/plane/generate.py`, `plane/tasks/{verify,agree,roles,repo_facts}.py`.
+  These are left for the owners of those packages.

benchmarks/archive/score_batched.py

@@ -10,7 +10,7 @@ detected when an AGREED finding matches; unstable when only a DISPUTED one does;
 finding at the fixed pin matches the fix's new side. The precision sample is every third agreed vulnerable-pin
 finding (all if 15 or fewer), to be adjudicated by hand. v2 is spent for tuning: this qualifies nothing.
 
-Usage: python benchmarks/independent/score_batched.py [--results-suffix batched] [--population population-v2.toml]
+Usage: python benchmarks/archive/score_batched.py [--results-suffix batched] [--population population-v2.toml]
 """
 
 from __future__ import annotations
@@ -20,7 +20,8 @@ import json
 import sys
 from pathlib import Path
 
-HERE = Path(__file__).resolve().parent
+# Archived 2026-10-02 from benchmarks/independent/: HERE stays that directory (population, evaluate.py, results).
+HERE = Path(__file__).resolve().parents[1] / "independent"
 sys.path.insert(0, str(HERE))
 
 import evaluate  # noqa: E402

benchmarks/archive/sink_candidates.py

@@ -12,7 +12,7 @@ changed files -- never to choose scope.
 
 v2 is spent for tuning; this informs the selection design and qualifies nothing.
 
-Usage: python benchmarks/independent/sink_candidates.py [--population population-v2.toml] [--only ID,...]
+Usage: python benchmarks/archive/sink_candidates.py [--population population-v2.toml] [--only ID,...]
 """
 
 from __future__ import annotations
@@ -24,7 +24,8 @@ import shutil
 import sys
 from pathlib import Path
 
-HERE = Path(__file__).resolve().parent
+# Archived 2026-10-02 from benchmarks/independent/: HERE stays that directory (population, evaluate.py, results).
+HERE = Path(__file__).resolve().parents[1] / "independent"
 sys.path.insert(0, str(HERE))
 
 import evaluate  # noqa: E402

benchmarks/archive/verify_batched.py

@@ -22,7 +22,7 @@ Source only; the stage-1 candidates come from `model_sinks.py`. Results in `~/ou
 EXPLORATORY: v2 is spent for tuning; this qualifies nothing.
 
 Usage:
-    python benchmarks/independent/verify_batched.py [--only ID,...] [--subset sets.json] [--passes 2] [--budget-usd 40]
+    python benchmarks/archive/verify_batched.py [--only ID,...] [--subset sets.json] [--passes 2] [--budget-usd 40]
 """
 
 from __future__ import annotations
@@ -35,8 +35,10 @@ import sys
 import time
 from pathlib import Path
 
-HERE = Path(__file__).resolve().parent
+# Archived 2026-10-02 from benchmarks/independent/: HERE stays that directory (population, evaluate.py, results).
+HERE = Path(__file__).resolve().parents[1] / "independent"
 sys.path.insert(0, str(HERE))
+sys.path.insert(0, str(Path(__file__).resolve().parent))  # archived siblings: model_sinks, verify_sinks
 
 import evaluate  # noqa: E402
 from model_sinks import OPERATIONS, chunks  # noqa: E402

benchmarks/archive/verify_sinks.py

@@ -10,7 +10,7 @@ Scope: the candidates `model_sinks.py` recorded for each case's vulnerable pin
 detected sites (`--fixed`). Findings are written in the scan-result shape `evaluate.py` reads. Hard spend
 ceiling. EXPLORATORY: v2 is spent for tuning; this qualifies nothing.
 
-Usage: python benchmarks/independent/verify_sinks.py [--only ID,...] [--budget-usd 40] [--fixed]
+Usage: python benchmarks/archive/verify_sinks.py [--only ID,...] [--budget-usd 40] [--fixed]
 """
 
 from __future__ import annotations
@@ -23,8 +23,10 @@ import sys
 import time
 from pathlib import Path
 
-HERE = Path(__file__).resolve().parent
+# Archived 2026-10-02 from benchmarks/independent/: HERE stays that directory (population, evaluate.py, results).
+HERE = Path(__file__).resolve().parents[1] / "independent"
 sys.path.insert(0, str(HERE))
+sys.path.insert(0, str(Path(__file__).resolve().parent))  # archived siblings: model_sinks, verify_sinks
 
 import evaluate  # noqa: E402
 from model_sinks import OPERATIONS  # noqa: E402

tests/test_harnessx_removal_equality.py

@@ -12,7 +12,7 @@ import json
 from pathlib import Path
 
 ROOT = Path(__file__).resolve().parents[1]
-_SPEC = importlib.util.spec_from_file_location("hxr_equality", ROOT / "benchmarks" / "harnessx_removal_equality.py")
+_SPEC = importlib.util.spec_from_file_location("hxr_equality", ROOT / "benchmarks" / "archive" / "harnessx_removal_equality.py")
 assert _SPEC is not None and _SPEC.loader is not None
 equality = importlib.util.module_from_spec(_SPEC)
 _SPEC.loader.exec_module(equality)

tests/test_plane_scoring_parity.py

@@ -1,6 +1,6 @@
 """`agree` scores exactly as the reference measurement did (ai-service-plane Requirement 6).
 
-The reference is `benchmarks/independent/score_batched.py` over the batched-check scans: sites anchored by
+The reference is `benchmarks/archive/score_batched.py` over the batched-check scans: sites anchored by
 `score_batched.anchored`, matched by `evaluate.matches_v2` against `evaluate.hunks(case, "old")`. The plane cannot
 ship the benchmark tree, so the generator resolves the ranges into `case.json` and `agree.matches_declared`
 reproduces the rule. These tests run both on the recorded reference outputs of the 15 validation cases and
@@ -23,6 +23,7 @@ from openultrasast.plane.manifests import load_manifests
 from openultrasast.plane.tasks import agree
 
 BENCH = Path("benchmarks/independent")
+ARCHIVE = Path("benchmarks/archive")  # score_batched.py, archived 2026-10-02
 POPULATION = BENCH / "population-v2.toml"
 SCANS = Path.home() / "ousast-results" / "independent-v2-batched-check" / "scans"
 REPOS = Path.home() / ".cache" / "openultrasast" / "independent"
@@ -33,10 +34,10 @@ pytestmark = pytest.mark.skipif(
 )
 
 
-def _load(name: str) -> ModuleType:
+def _load(name: str, where: Path = BENCH) -> ModuleType:
     sys.path.insert(0, str(BENCH.resolve()))  # score_batched imports evaluate by name
     try:
-        spec = importlib.util.spec_from_file_location(f"parity_{name}", BENCH / f"{name}.py")
+        spec = importlib.util.spec_from_file_location(f"parity_{name}", where / f"{name}.py")
         assert spec is not None and spec.loader is not None
         module = importlib.util.module_from_spec(spec)
         spec.loader.exec_module(module)
@@ -47,7 +48,7 @@ def _load(name: str) -> ModuleType:
 
 @pytest.fixture(scope="module")
 def reference() -> tuple[ModuleType, ModuleType]:
-    evaluate, score_batched = _load("evaluate"), _load("score_batched")
+    evaluate, score_batched = _load("evaluate"), _load("score_batched", ARCHIVE)
     evaluate.use(POPULATION.resolve())
     return evaluate, score_batched