fix(ci): ruff format on verification harnesses + bump tool count to 47

- 12 files in benchmarks/lib/* + tool_registry_memory.py reformatted to
  pass `ruff format --check .` on CI
- tests_py/test_main.py: tool_names now contains 47 entries after the
  paper+verification commit added get_telemetry (Popper C6 — read/write
  ratio measurement). Updated assertion + added explicit get_telemetry
  presence check

Validates locally: ruff format --check on source tree clean (572 files);
test_mcp_server_has_tools passes.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: cdeust

benchmarks/lib/_xb_drivers.py

@@ -23,6 +23,7 @@
 
 def _drive_longmemeval(data_path: str, limit: int) -> dict:
     from benchmarks.longmemeval.run_benchmark import run_benchmark
+
     r = run_benchmark(data_path, limit=limit)
     return {
         "mrr": r["overall_mrr"],
@@ -67,7 +68,9 @@ def _drive_locomo(data_path: str, limit: int) -> dict:
     n = len(all_rs)
     mrr = sum(1.0 / r["hit_rank"] for r in all_rs if r["hit_rank"]) / n if n else 0.0
     r10 = (
-        sum(1 for r in all_rs if r["hit_rank"] and r["hit_rank"] <= 10) / n if n else 0.0
+        sum(1 for r in all_rs if r["hit_rank"] and r["hit_rank"] <= 10) / n
+        if n
+        else 0.0
     )
     cat_mrr: dict[str, float] = {}
     cat_r10: dict[str, float] = {}
@@ -76,9 +79,7 @@ def _drive_locomo(data_path: str, limit: int) -> dict:
             continue
         m = len(rs)
         cat_mrr[cat] = sum(1.0 / r["hit_rank"] for r in rs if r["hit_rank"]) / m
-        cat_r10[cat] = sum(
-            1 for r in rs if r["hit_rank"] and r["hit_rank"] <= 10
-        ) / m
+        cat_r10[cat] = sum(1 for r in rs if r["hit_rank"] and r["hit_rank"] <= 10) / m
     return {
         "mrr": mrr,
         "recall_at_10": r10,
@@ -91,8 +92,10 @@ def _drive_locomo(data_path: str, limit: int) -> dict:
 
 def main() -> int:
     if len(sys.argv) != 4:
-        print("usage: _xb_drivers.py <longmemeval|locomo> <data_path> <limit>",
-              file=sys.stderr)
+        print(
+            "usage: _xb_drivers.py <longmemeval|locomo> <data_path> <limit>",
+            file=sys.stderr,
+        )
         return 2
     bench, data_path, limit_s = sys.argv[1], sys.argv[2], sys.argv[3]
     limit = int(limit_s)

benchmarks/lib/ablation_runner.py

@@ -31,7 +31,8 @@
 from mcp_server.core import ablation as _ablation  # noqa: E402
 from mcp_server.core.ablation import Mechanism  # noqa: E402
 from benchmarks.lib.db_snapshot import (  # noqa: E402
-    fingerprint, restore_snapshot,
+    fingerprint,
+    restore_snapshot,
 )
 from benchmarks.lib import db_setup  # noqa: E402
 
@@ -73,16 +74,18 @@ def _patches_for(mech: Mechanism) -> list[PatchSpec]:
     no_decay: PatchSpec = (
         "mcp_server.core.thermodynamics",
         "compute_decay",
-        lambda: (lambda current_heat, *a, **k: current_heat),
+        lambda: lambda current_heat, *a, **k: current_heat,
     )
     table: dict[Mechanism, list[PatchSpec]] = {
         Mechanism.OSCILLATORY_CLOCK: [no_decay],
         Mechanism.ADAPTIVE_DECAY: [no_decay],
-        Mechanism.HOMEOSTATIC_PLASTICITY: [(
-            "mcp_server.core.ablation",
-            "neutral_scaling_factor",
-            lambda: _ablation.neutral_scaling_factor,
-        )],
+        Mechanism.HOMEOSTATIC_PLASTICITY: [
+            (
+                "mcp_server.core.ablation",
+                "neutral_scaling_factor",
+                lambda: _ablation.neutral_scaling_factor,
+            )
+        ],
     }
     return table.get(mech, [])
 
@@ -126,11 +129,11 @@ def _parse_metrics(stdout: str) -> BenchMetrics:
     """
     r10 = mrr = 0.0
     n = 0
-    if (m := _RE_R10.search(stdout)):
+    if m := _RE_R10.search(stdout):
         r10 = float(m.group(1)) / 100.0
-    if (m := _RE_MRR_LINE.search(stdout)):
+    if m := _RE_MRR_LINE.search(stdout):
         mrr = float(m.group(1))
-    if (m := _RE_OVERALL.search(stdout)):
+    if m := _RE_OVERALL.search(stdout):
         mrr = float(m.group(1))
         r10 = float(m.group(3)) / 100.0
         n = int(m.group(4))
@@ -146,16 +149,19 @@ def _resolve_benchmark(bench_id: str, quick: bool) -> tuple[Callable[[], None],
     """Return (callable that runs the benchmark, declared n_queries-or-0)."""
     if bench_id == "longmemeval-s":
         from benchmarks.longmemeval.run_benchmark import run_benchmark as r
+
         data_path = _ROOT / "benchmarks" / "longmemeval" / "longmemeval_s.json"
         limit = QUICK_LIMIT if quick else 0
         return (lambda: r(str(data_path), limit=limit, verbose=False), limit)
     if bench_id == "locomo":
         from benchmarks.locomo.run_benchmark import run_benchmark as r
+
         data_path = _ROOT / "benchmarks" / "locomo" / "locomo10.json"
         limit = max(1, QUICK_LIMIT // 10) if quick else None
         return (lambda: r(str(data_path), limit=limit, verbose=False), limit or 0)
     if bench_id == "beam-100K":
         from benchmarks.beam.run_benchmark import run_benchmark as r
+
         limit = max(1, QUICK_LIMIT // 10) if quick else None
         return (lambda: r(split="100K", limit=limit, verbose=False), limit or 0)
     raise ValueError(f"unknown benchmark: {bench_id}")
@@ -310,19 +316,23 @@ def _select_mechanisms(args: argparse.Namespace) -> list[Mechanism]:
 
 
 def _maybe_restore(
-    snapshot_path: Path | None, target_db_url: str | None,
-    *, run_id: str = "ablation",
+    snapshot_path: Path | None,
+    target_db_url: str | None,
+    *,
+    run_id: str = "ablation",
 ) -> dict | str:
     """Restore snapshot (if any), apply deterministic DB+session config,
     ANALYZE, return db_seed payload (playbook §4.3, §4.10, §4.15)."""
     if snapshot_path is None or target_db_url is None:
         return "not-snapshotted"
     sha = fingerprint(snapshot_path)
-    print(f"  [snapshot] restoring {snapshot_path.name} (sha8={sha[:8]}) -> "
-          f"{target_db_url} ...")
+    print(
+        f"  [snapshot] restoring {snapshot_path.name} (sha8={sha[:8]}) -> "
+        f"{target_db_url} ..."
+    )
     report = restore_snapshot(target_db_url, snapshot_path)
     if not report.success:
-        msg = '; '.join(report.mismatch + report.version_drift)
+        msg = "; ".join(report.mismatch + report.version_drift)
         raise RuntimeError(f"snapshot restore failed: {msg}")
     os.environ["DATABASE_URL"] = target_db_url
     # Post-restore deterministic setup (playbook §8 SRP split).
@@ -348,8 +358,11 @@ def _maybe_restore(
 
 
 def _run_baseline_if_needed(
-    benchmark: str, quick: bool, force: bool,
-    snapshot: Path | None, target_db: str | None,
+    benchmark: str,
+    quick: bool,
+    force: bool,
+    snapshot: Path | None,
+    target_db: str | None,
 ) -> BenchMetrics:
     cached = None if force else _load_baseline(benchmark)
     if cached is not None:
@@ -358,8 +371,9 @@ def _run_baseline_if_needed(
     print(f"  [baseline] running ({benchmark}, quick={quick}) ...")
     run_id = f"{benchmark}_BASELINE"
     db_seed = _maybe_restore(snapshot, target_db, run_id=run_id)
-    metrics, wall, rss, _ = run_one(benchmark, None, quick=quick,
-                                    run_id=run_id if snapshot else None)
+    metrics, wall, rss, _ = run_one(
+        benchmark, None, quick=quick, run_id=run_id if snapshot else None
+    )
     _save(benchmark, None, metrics, None, wall, rss, db_seed=db_seed)
     print(
         f"  [baseline] r@10={metrics.r_at_10:.3f} mrr={metrics.mrr:.3f} "
@@ -376,26 +390,43 @@ def main() -> int:
     p.add_argument("--baseline-only", action="store_true")
     p.add_argument("--quick", action="store_true")
     p.add_argument("--force", action="store_true")
-    p.add_argument("--from-snapshot", dest="from_snapshot", default=None,
-                   help="path to a custom-format pg_dump; restored before each trial")
-    p.add_argument("--target-db", dest="target_db", default=None,
-                   help="benchmark-only DB URL/name to restore the snapshot into")
+    p.add_argument(
+        "--from-snapshot",
+        dest="from_snapshot",
+        default=None,
+        help="path to a custom-format pg_dump; restored before each trial",
+    )
+    p.add_argument(
+        "--target-db",
+        dest="target_db",
+        default=None,
+        help="benchmark-only DB URL/name to restore the snapshot into",
+    )
     args = p.parse_args()
 
     snapshot = Path(args.from_snapshot) if args.from_snapshot else None
     target_db = None
     if snapshot is not None:
         if not args.target_db:
             raise SystemExit("--from-snapshot requires --target-db")
-        target_db = (args.target_db if "://" in args.target_db
-                     else f"postgresql://localhost:5432/{args.target_db}")
+        target_db = (
+            args.target_db
+            if "://" in args.target_db
+            else f"postgresql://localhost:5432/{args.target_db}"
+        )
     else:
-        print("WARNING: --from-snapshot not set; results lack DB-state determinism. "
-              "HNSW non-determinism + ingest order will leak into the deltas.")
+        print(
+            "WARNING: --from-snapshot not set; results lack DB-state determinism. "
+            "HNSW non-determinism + ingest order will leak into the deltas."
+        )
 
     print(f"== ablation runner :: benchmark={args.benchmark} quick={args.quick} ==")
     baseline = _run_baseline_if_needed(
-        args.benchmark, args.quick, args.force, snapshot, target_db,
+        args.benchmark,
+        args.quick,
+        args.force,
+        snapshot,
+        target_db,
     )
     if args.baseline_only:
         return 0
@@ -413,9 +444,12 @@ def main() -> int:
         try:
             run_id = f"{args.benchmark}_{mech.name}"
             db_seed = _maybe_restore(snapshot, target_db, run_id=run_id)
-            metrics, wall, rss, _ = run_one(args.benchmark, mech,
-                                            quick=args.quick,
-                                            run_id=run_id if snapshot else None)
+            metrics, wall, rss, _ = run_one(
+                args.benchmark,
+                mech,
+                quick=args.quick,
+                run_id=run_id if snapshot else None,
+            )
         except Exception as exc:  # source: ablation must be fail-soft per mech
             print(f"  [{mech.name}] FAILED: {exc!r}")
             continue

benchmarks/lib/bench_db.py

@@ -32,8 +32,10 @@ class BenchmarkDB:
     """
 
     def __init__(
-        self, database_url: str | None = None,
-        *, on_connection_open: Callable[[Any], None] | None = None,
+        self,
+        database_url: str | None = None,
+        *,
+        on_connection_open: Callable[[Any], None] | None = None,
     ) -> None:
         self._url = database_url or os.environ.get(
             "DATABASE_URL", "postgresql://localhost:5432/cortex"
@@ -54,6 +56,7 @@ def open(self) -> BenchmarkDB:
         run_id = os.environ.get("CORTEX_BENCH_DETERMINISTIC_RUN_ID")
         if run_id and self._on_connection_open is None:
             from benchmarks.lib import db_setup
+
             db_setup.apply_deterministic_session(self._store._conn, run_id=run_id)
         elif self._on_connection_open is not None:
             self._on_connection_open(self._store._conn)