Harden detection model asset selection

CHANGELOG.md

@@ -1292,3 +1292,12 @@ Added:
 - Observed dense QA F1 scores up to `0.6380` and safest current threshold behavior around `0.25`.
 - Kept the model inactive by default because the `kasterlee_bos` hard-negative sample still produced 10 detections at threshold `0.25`.
 - No repository code, API contracts, migrations, product behavior, provider fetching or AI dependency strategy changed in this benchmark pass.
+
+## Sprint 122 Detection model asset activation guardrails (2026-07-08)
+
+- Hardened Detection Lab so local model assets are no longer auto-selected when the backend reports available model files.
+- Required an explicit local model asset choice before configured YOLO can be submitted when local assets exist.
+- Added local model asset details in the run surface: active runtime env status, SHA-256 preview, file size, path and `will_download_models`.
+- Surfaced the current YOLOv8s hard-negative benchmark candidate and recommended starting threshold `0.25` as operator guidance.
+- Added regression coverage for the no-auto-select behavior and UI guardrail copy.
+- No backend API contracts, migrations, model downloads, provider fetching or model weight mutation behavior changed.

backend/tests/test_sprint122_model_asset_activation_guardrails.py

@@ -0,0 +1,37 @@
+from pathlib import Path
+
+
+ROOT = Path(__file__).resolve().parents[2]
+
+
+def test_detection_workflow_does_not_auto_select_runtime_model_assets() -> None:
+    hook = (ROOT / "frontend" / "src" / "hooks" / "useDetectionWorkflow.ts").read_text(encoding="utf-8")
+
+    assert "const activeAsset = assetResponse.items.find" not in hook
+    assert "setSelectedModelAssetId(activeAsset?.model_asset_id ?? '')" not in hook
+    assert "setSelectedModelAssetId(assetResponse.items[0]" not in hook
+    assert "setSelectedModelAssetId('')" in hook
+
+
+def test_detection_lab_explains_explicit_model_asset_and_threshold_selection() -> None:
+    lab = (ROOT / "frontend" / "src" / "components" / "detection" / "DetectionLab.tsx").read_text(
+        encoding="utf-8"
+    )
+
+    assert "Explicit model asset" in lab
+    assert "No model file is selected automatically" in lab
+    assert "Current benchmark candidate" in lab
+    assert "geointel-building-yolov8s-hardneg160r4e50-pt" in lab
+    assert "Recommended starting threshold" in lab
+    assert "0.25" in lab
+    assert "will_download_models" in lab
+
+
+def test_detection_run_readiness_requires_explicit_asset_when_local_assets_exist() -> None:
+    lab = (ROOT / "frontend" / "src" / "components" / "detection" / "DetectionLab.tsx").read_text(
+        encoding="utf-8"
+    )
+
+    assert "detectionHasExplicitModelAsset" in lab
+    assert "Select a local model asset deliberately" in lab
+    assert "local model asset" in lab

docs/CODEX_EXECUTION_LOG.md

@@ -5000,6 +5000,36 @@ Limitations:
 Next recommended pass:
 - Continue with V1 usability work that reduces operator confusion without expanding frozen product scope.
 
+## Sprint 122 Detection model asset activation guardrails (2026-07-08)
+
+Changed:
+- Hardened Detection Lab local model handling so reported runtime model assets are read-only choices and are not auto-selected by the frontend hook.
+- Configured YOLO run readiness now blocks submission when local model assets exist but no explicit `model_asset_id` has been selected.
+- Added an explicit model asset section with active runtime env status, `will_download_models`, SHA-256 preview, file size and mounted model path.
+- Surfaced the current benchmark candidate `geointel-building-yolov8s-hardneg160r4e50-pt` with recommended starting threshold `0.25`.
+- Added compact UI guidance styling for the benchmark/threshold warning.
+- Added regression coverage in `backend/tests/test_sprint122_model_asset_activation_guardrails.py`.
+- Updated `CHANGELOG.md` and `docs/TODO.md`.
+
+Tested:
+- Red step: `python -m pytest backend\tests\test_sprint122_model_asset_activation_guardrails.py -q` failed on the previous auto-selection behavior and missing guardrail copy.
+- `python -m pytest backend\tests\test_sprint122_model_asset_activation_guardrails.py -q` (`3 passed`)
+- `python -m pytest backend\tests\test_sprint118_yolo_preflight_ui.py backend\tests\test_sprint103_ai_lab_run_readiness.py backend\tests\test_sprint104_ai_lab_action_guardrails.py backend\tests\test_model_asset_catalog.py backend\tests\test_sprint122_model_asset_activation_guardrails.py -q` (`16 passed`)
+- `python -m compileall backend/app`
+- `python -m pytest backend\tests -q` (`408 passed`)
+- `cd frontend && npm run typecheck`
+- `cd frontend && npm run build`
+
+Open:
+- Full threshold calibration comparison UI is still pending; this pass adds safe single-threshold guidance and explicit asset choice only.
+
+Limitations:
+- Frontend guardrail only; no backend API contracts, migrations, provider fetching, model downloads or model weight mutation behavior changed.
+- The active runtime env model can still be configured by operators through existing deployment/env tooling, but the Detection Lab no longer silently chooses a local asset from the catalog for a run.
+
+Next recommended pass:
+- Add threshold calibration comparison UX over existing persisted runs so candidate models can be promoted with visible precision/recall/F1 and hard-negative counts.
+
 ## Operator YOLOv8s hard-negative model benchmark (2026-07-08)
 
 Changed:

docs/TODO.md

@@ -416,7 +416,8 @@ This file now starts with the current implementation status. Older preparation/b
 - [x] Add operator-only tile-level YOLO dataset export with overlapping windows and deterministic negative tile retention.
 - [x] Run tile-level training on Tower and accept/reject the resulting local model through the persisted QA/QC matrix.
 - [x] Train/evaluate a YOLOv8s hard-negative local building-detector candidate on Tower and keep it inactive because hard-negative false positives remain.
-- [ ] Add an operator-facing local model catalog/activation workflow with SHA256, active model status and explicit threshold guidance.
-- [ ] Add threshold calibration UX so detection runs do not silently rely on an unsafe default confidence.
+- [x] Add an operator-facing local model catalog/activation workflow with SHA256, active model status and explicit threshold guidance.
+- [x] Block silent local model asset auto-selection in Detection Lab.
+- [ ] Add full threshold calibration comparison UX so detection runs can compare candidate thresholds before promotion.
 - [ ] Add more AOIs after the tile-level baseline so the next local model attempt is not limited to Geel/Mol/Turnhout.
 - [ ] Add negative/background AOIs so the next tile dataset is not all positive tiles.

frontend/src/components/detection/DetectionLab.tsx

@@ -109,10 +109,20 @@ export function DetectionLab({
   const detectionHasModel = selectedDetectionModel !== null
   const detectionModelReady = Boolean(selectedDetectionModel?.configured)
   const detectionModelUiRunnable = detectionModelReady && selectedDetectionModelId !== 'manual-fixture-detector'
+  const detectionHasExplicitModelAsset =
+    selectedDetectionModelId !== 'yolo-configured' || modelAssets.length === 0 || selectedModelAssetId.length > 0
+  const benchmarkCandidateAsset = modelAssets.find(
+    (asset) => asset.model_asset_id === 'geointel-building-yolov8s-hardneg160r4e50-pt',
+  )
   const detectionHasTileManifest =
     !detectionRequiresTileManifest || detectionTileManifestPath.trim().length > 0
   const detectionRunReady =
-    Boolean(selectedProjectId) && detectionHasDataset && detectionHasModel && detectionModelUiRunnable && detectionHasTileManifest
+    Boolean(selectedProjectId) &&
+    detectionHasDataset &&
+    detectionHasModel &&
+    detectionModelUiRunnable &&
+    detectionHasExplicitModelAsset &&
+    detectionHasTileManifest
   const detectionRunBlockedReason = !selectedProjectId
     ? 'Select or create a project first'
     : !detectionHasDataset
@@ -123,6 +133,8 @@ export function DetectionLab({
           ? 'Fixture model is explicit test/demo-only'
           : !detectionModelReady
             ? selectedDetectionModel?.limitation_message ?? 'Selected model is not configured'
+            : !detectionHasExplicitModelAsset
+              ? 'Select a local model asset deliberately'
             : !detectionHasTileManifest
               ? 'Provide a raster tile manifest for configured YOLO'
               : null
@@ -193,17 +205,26 @@ export function DetectionLab({
         <div className="ai-lab-model-surface" aria-label="Local model asset selection">
           <div className="ai-lab-section-header">
             <div>
-              <h3>Local model assets</h3>
-              <p>Select an existing model file mounted into the backend runtime. GeoIntel does not download model weights.</p>
+              <h3>Explicit model asset</h3>
+              <p>Local model assets are read-only. No model file is selected automatically. Choose an existing local model asset before submitting configured YOLO.</p>
             </div>
             <span className={selectedModelAsset ? 'status-badge status-badge-ready' : 'status-badge'}>
-              {selectedModelAsset ? 'asset selected' : 'using configured path'}
+              {selectedModelAsset ? 'asset selected' : 'no explicit asset'}
             </span>
           </div>
+          {benchmarkCandidateAsset ? (
+            <div className="model-asset-guidance">
+              <strong>Current benchmark candidate</strong>
+              <p>
+                {benchmarkCandidateAsset.display_name} is available for deliberate evaluation. Recommended starting threshold: 0.25.
+                Keep it operator-selected until hard-negative false positives are reduced.
+              </p>
+            </div>
+          ) : null}
           <label>
             Local model file
             <select value={selectedModelAssetId} onChange={(event) => onSelectModelAsset(event.target.value)}>
-              <option value="">Use configured YOLO_MODEL_PATH</option>
+              <option value="">Select explicit local model asset</option>
               {modelAssets.map((asset) => (
                 <option key={asset.model_asset_id} value={asset.model_asset_id}>
                   {asset.display_name} {asset.active ? '(active)' : ''}
@@ -221,6 +242,8 @@ export function DetectionLab({
             <div className="result-summary-card">
               <p>File: {selectedModelAsset.filename}</p>
               <p>Status: {selectedModelAsset.status}</p>
+              <p>Active runtime env model: {selectedModelAsset.active ? 'yes' : 'no'}</p>
+              <p>will_download_models: {selectedModelAsset.will_download_models ? 'yes' : 'no'}</p>
               <p>Size: {formatModelAssetSize(selectedModelAsset.size_bytes)}</p>
               <p>SHA-256: {selectedModelAsset.sha256.slice(0, 12)}</p>
               <p>Path: {selectedModelAsset.model_path}</p>
@@ -340,6 +363,18 @@ export function DetectionLab({
                     : 'Not required for this model'}
                 </strong>
               </div>
+              <div className={detectionHasExplicitModelAsset ? 'lab-readiness-item lab-readiness-item-ready' : 'lab-readiness-item'}>
+                <span>Local model asset</span>
+                <strong>
+                  {selectedDetectionModelId === 'yolo-configured'
+                    ? selectedModelAsset
+                      ? selectedModelAsset.display_name
+                      : modelAssets.length > 0
+                        ? 'Select a local model asset deliberately'
+                        : 'No local assets reported; backend configured path only'
+                    : 'Not required for this model'}
+                </strong>
+              </div>
             </div>
           </div>
           <div className={detectionRunReady ? 'lab-action-guardrail lab-action-guardrail-ready' : 'lab-action-guardrail'}>
@@ -384,6 +419,11 @@ export function DetectionLab({
                 value={detectionConfidenceThreshold}
                 onChange={(event) => onSetConfidenceThreshold(Number(event.target.value))}
               />
+              {selectedDetectionModelId === 'yolo-configured' ? (
+                <span className="field-guidance">
+                  Recommended starting threshold: 0.25 for the current local YOLOv8s benchmark candidate.
+                </span>
+              ) : null}
             </label>
           </div>
           {selectedDetectionModelId === 'yolo-configured' ? (

frontend/src/hooks/useDetectionWorkflow.ts

@@ -72,9 +72,8 @@ export function useDetectionWorkflow({
     try {
       const assetResponse = await detectionApi.listModelAssets()
       setModelAssets(assetResponse.items)
-      const activeAsset = assetResponse.items.find((asset) => asset.active) ?? assetResponse.items[0] ?? null
       if (!assetResponse.items.some((asset) => asset.model_asset_id === selectedModelAssetId)) {
-        setSelectedModelAssetId(activeAsset?.model_asset_id ?? '')
+        setSelectedModelAssetId('')
       }
     } catch (error) {
       setModelAssets([])

frontend/src/styles/app.css

@@ -3369,6 +3369,34 @@ button.entity-card {
   overflow-wrap: anywhere;
 }
 
+.model-asset-guidance {
+  display: grid;
+  gap: 0.22rem;
+  border: 1px solid #d8e3de;
+  border-left: 4px solid #315a92;
+  border-radius: 8px;
+  padding: 0.58rem 0.68rem;
+  background: #f8fbff;
+}
+
+.model-asset-guidance strong {
+  color: var(--text);
+  font-size: 0.9rem;
+}
+
+.model-asset-guidance p,
+.field-guidance {
+  margin: 0;
+  color: var(--muted);
+  font-size: 0.78rem;
+  line-height: 1.35;
+}
+
+.field-guidance {
+  display: block;
+  margin-top: 0.24rem;
+}
+
 .lab-action-guardrail {
   display: flex;
   min-width: 0;