refactor(metrics): delete the metrics journal and model registry (XWING-2440)

.ratchet.json

@@ -25,12 +25,6 @@
     "contracts": 1,
     "state-file": 4
   },
-  "src/traust_engine/metrics/history.py": {
-    "contracts": 2
-  },
-  "src/traust_engine/metrics/spend.py": {
-    "contracts": 1
-  },
   "src/traust_engine/ops/admission.py": {
     "core-internal": 2
   },
@@ -46,12 +40,6 @@
   "src/traust_engine/portfolio/graph.py": {
     "database": 1
   },
-  "src/traust_engine/registry/models.py": {
-    "contracts": 1
-  },
-  "src/traust_engine/registry/ops.py": {
-    "contracts": 1
-  },
   "src/traust_engine/sweep/engine.py": {
     "contracts": 1
   },

CODEOWNERS

@@ -10,7 +10,4 @@
 # Disposition ledger and corpus resolution
 /src/traust_engine/corpus/ @traust-security/traust-admins
 
-# Metrics journal
-/src/traust_engine/metrics/history.py @traust-security/traust-admins
-
 * @traust-security/traust-admins

README.md

@@ -15,18 +15,17 @@ tagged `PENDING(decision: …)`); deleted when traust retires or ports it:
 
 | Module | Kept for |
 |---|---|
-| `traust_engine.engine` | `HarnessEngine` facade: `.metrics`, `.models`, `.adapters`, `.sweep`, `.portfolio` |
+| `traust_engine.engine` | `HarnessEngine` facade: `.adapters`, `.sweep`, `.portfolio` |
 | `traust_engine.sweep` | sweep stages: collect, draft, sweep, emit, benchmark, rule lane, mining, calibration |
 | `traust_engine.portfolio` | portfolio graph layers |
-| `traust_engine.metrics`, `traust_engine.registry` | metrics journal and spend pricing |
 | `traust_engine.adapters` | opengrep (sweep stages) and checkov (portfolio graph) |
 | `traust_engine.corpus` | the report population the sweep stages read |
 
 The opengrep rule pack ships under `data/secure-code-audit/opengrep-rules/`. This library ships **no configuration files and never loads
 config itself — it is a pure consumer.** A caller builds the context once
 (`traust_contracts.load_context()`, done for you by
 `HarnessEngine.load(config_home=…)`) and the engine receives it *injected*: the
-ops namespaces (`h.metrics`, `h.adapters`, …) hold the frozen `HarnessContext` and
+ops namespaces (`h.sweep`, `h.adapters`, …) hold the frozen `HarnessContext` and
 derive paths from it; stateless functions take already-resolved arguments. No
 environment overrides, no cwd or workspace-shaped guessing; a required-but-absent
 file fails closed with `DeploymentConfigMissing` at the entry point. The config

pyproject.toml

@@ -63,4 +63,3 @@ ignore = ["RUF001", "RUF002", "RUF003"]
 
 [tool.ruff.lint.per-file-ignores]
 "tests/*" = ["S101", "E741", "E402", "E501", "RUF012"]
-"src/traust_engine/metrics/history.py" = ["E501"]

src/traust_engine/__init__.py

@@ -4,12 +4,10 @@
     ops         — continuous ops: pack spec, controller (tick), admission, reaper
 
 Kept for traust until its PENDING decisions (sweep stages, portfolio graph
-layers, metrics journal) retire or port them:
-    engine      — the ``HarnessEngine`` facade (metrics, models, adapters, sweep, portfolio)
+layers) retire or port them:
+    engine      — the ``HarnessEngine`` facade (adapters, sweep, portfolio)
     sweep       — class-generalization scan engine
     portfolio   — source-code dependency graph analysis
-    metrics     — hash-chained metrics recording + spend tracking
-    registry    — model resolution and spend pricing
     adapters    — opengrep (sweep stages) and checkov (portfolio graph)
     corpus      — population resolver the sweep stages read
 """

src/traust_engine/_ops_base.py

@@ -4,7 +4,7 @@
 needs. The resolution logic itself lives in :mod:`traust_engine.locations`; this
 just binds it to ``self._ctx.locations`` so no namespace re-derives it.
 
-PENDING(decision: sweep stages | portfolio graph layers | metrics journal) — kept for traust;
+PENDING(decision: sweep stages | portfolio graph layers) — kept for traust;
 deleted when traust retires or ports it.
 """
 

src/traust_engine/assets.py

@@ -9,7 +9,7 @@
 arithmetic or a repo-layout assumption — the latter breaks the moment the
 package is installed as a wheel (the original HARNESS_ROOT bug).
 
-PENDING(decision: sweep stages | metrics journal) — kept for traust;
+PENDING(decision: sweep stages) — kept for traust;
 deleted when traust retires or ports it.
 """
 

src/traust_engine/engine.py

@@ -8,13 +8,12 @@
     from traust_engine import HarnessEngine
 
     h = HarnessEngine(ctx)                       # explicit context (tests, app)
-    h.metrics.append("census", {"repos": 812})
-    ok, issues = h.metrics.verify()
+    h.sweep.collect(...)
 
-Namespaces mirror the engine's submodules: `.metrics`, `.models`, `.adapters` (the rule-lane
-stage's opengrep config), `.sweep`, `.portfolio`.
+Namespaces mirror the engine's submodules: `.adapters` (the rule-lane stage's opengrep
+config), `.sweep`, `.portfolio`.
 
-PENDING(decision: sweep stages | portfolio graph layers | metrics journal) — kept for traust;
+PENDING(decision: sweep stages | portfolio graph layers) — kept for traust;
 deleted when traust retires or ports it.
 """
 
@@ -26,9 +25,7 @@
 from traust_contracts import HarnessContext, load_context
 
 from traust_engine.adapters.ops import AdaptersOps
-from traust_engine.metrics.ops import MetricsOps
 from traust_engine.portfolio.ops import PortfolioOps
-from traust_engine.registry.ops import ModelsOps
 from traust_engine.sweep.ops import SweepOps
 
 
@@ -45,14 +42,6 @@ def load(cls, *, config_home: Path | None = None) -> HarnessEngine:
     def ctx(self) -> HarnessContext:
         return self._ctx
 
-    @cached_property
-    def metrics(self) -> MetricsOps:
-        return MetricsOps(self._ctx)
-
-    @cached_property
-    def models(self) -> ModelsOps:
-        return ModelsOps(self._ctx)
-
     @cached_property
     def adapters(self) -> AdaptersOps:
         return AdaptersOps(self._ctx)

src/traust_engine/locations.py

@@ -11,7 +11,7 @@
 below the ops layer. The resolution logic lives here, once; ops pass ``self._loc``
 and never re-derive it.
 
-PENDING(decision: sweep stages | portfolio graph layers | metrics journal) — kept for traust;
+PENDING(decision: sweep stages | portfolio graph layers) — kept for traust;
 deleted when traust retires or ports it.
 """
 
@@ -141,8 +141,6 @@ def sarif_tool_uri(loc: Locations | None) -> str | None:
 PHASE0_REL = SCAN_TESTING_REL / "phase-0"
 SWEEPS_REL = SCAN_TESTING_REL / "sweeps"
 VALIDATION_BENCHMARK_REL = SCAN_TESTING_REL / "validation-benchmark"
-#: The metrics-history journal lives under the configured metrics root, not results.
-METRICS_HISTORY_REL = Path("metrics") / "metrics-history.jsonl"
 #: Sweep-engine and regression-rule draft staging under the metrics root.
 RULE_DRAFTS_REL = Path("metrics") / "rule-mining" / "rule-drafts"
 
@@ -197,9 +195,3 @@ def scan_testing_dir(loc: Locations | None) -> Path | None:
 def benchmark_dir(loc: Locations | None) -> Path | None:
     """``<analysis-results>/scan-testing/validation-benchmark``, or None."""
     return _under_results(VALIDATION_BENCHMARK_REL, loc)
-
-
-def metrics_history(loc: Locations | None) -> Path | None:
-    """``<metrics-root>/metrics/metrics-history.jsonl``, or None."""
-    pt = progress_tracker_dir(loc)
-    return pt / METRICS_HISTORY_REL if pt else None