Preroll udpate

shaders/SHADER_CONTRACT.md

@@ -105,6 +105,13 @@ Optional fields:
 
 Parameter objects may also include an optional `description` string. The control UI displays it as one-line helper text with the full text available on hover, so use it for short operational guidance rather than long documentation.
 
+Metadata conventions:
+
+- Keep `name` short, human-facing, and in title case.
+- Keep `category` consistent with existing library groups such as `Color`, `Transform`, `Projection`, `Temporal`, `Scopes & Guides`, `Utility`, `Feedback`, and `Calibration`.
+- Keep `description` to one clear sentence in present tense that explains what the shader does for an operator.
+- Avoid placeholder, joke, or overly implementation-heavy descriptions unless the shader is intentionally a diagnostic or broken example.
+
 Shader-visible identifiers must be valid Slang-style identifiers:
 
 - `entryPoint`
@@ -276,13 +283,15 @@ On the first frame, or after a reset, `sampleFeedback` returns transparent black
 
 Feedback resets when:
 
-- layers are added, removed, or reordered
 - a layer bypass state changes
 - a layer changes shader
+- the layer itself is removed
 - a shader is reloaded or recompiled
 - render dimensions change
 - the app restarts
 
+Ordinary stack add/remove/reorder operations on other layers are intended to preserve feedback state for unchanged feedback-enabled layers.
+
 So feedback should be treated as live runtime state, not durable saved state.
 
 ## Slang Entry Point

shaders/feedback-data-blocks/shader.json

@@ -1,7 +1,7 @@
 {
   "id": "feedback-data-blocks",
   "name": "Feedback Data Blocks",
-  "description": "Demonstrates using the feedback surface as coarse data storage by reserving eight 3x3 texel cells for sampled colors and a hidden metadata cell for timed or trigger-driven refresh state.",
+  "description": "Demonstrates coarse shader-local data storage by reserving eight 3x3 feedback cells for sampled colors and one hidden metadata cell for refresh state.",
   "category": "Feedback",
   "entryPoint": "storeProbeData",
   "passes": [

shaders/feedback-highlight-accumulator/shader.json

@@ -1,7 +1,7 @@
 {
   "id": "feedback-highlight-accumulator",
   "name": "Feedback Background Memory",
-  "description": "Demonstrates writable full-frame shader feedback by learning a persistent per-pixel background model, then comparing the live frame against that learned plate. This cannot be reproduced by only reading ordinary history frames because the shader writes its own evolving state back each frame.",
+  "description": "Learns a persistent per-pixel background plate in shader-local feedback and compares the live frame against that evolving full-frame state.",
   "category": "Feedback",
   "entryPoint": "updateBackgroundModel",
   "passes": [

shaders/white-balance-correction/shader.json

@@ -1,7 +1,7 @@
 {
   "id": "white-balance-correction",
   "name": "White Balance Correction",
-  "description": "Operator-friendly tint, color balance, and exposure correction intended to pair with the white match probe.",
+  "description": "Provides operator-friendly warm/cool, green/magenta, and exposure correction intended to pair with the White Match Probe.",
   "category": "Color",
   "entryPoint": "shadeVideo",
   "parameters": [

shaders/white-match-probe/shader.json

@@ -1,8 +1,8 @@
 {
   "id": "white-match-probe",
   "name": "White Match Probe",
-  "description": "Samples a movable box, stores a reference color on trigger using shader-local feedback, and compares the current sample against the held reference for camera matching.",
-  "category": "Utility",
+  "description": "Samples a movable box, stores a reference color on trigger using shader-local feedback, and compares the current sample against a captured or manual reference for camera matching.",
+  "category": "Scopes & Guides",
   "entryPoint": "storeReferenceState",
   "passes": [
     {