Fix encoder readiness and bundle workflow authoring repairs

data/research/comfy-contract-resolution.v1.json

@@ -13,7 +13,7 @@
     "publishesTemplates": false,
     "researchOnly": true
   },
-  "contentHash": "sha256:425d20ad21a3c8b86d66d7e254f65be47a98999758f5ffdf71e0ea79c8725214",
+  "contentHash": "sha256:ed9cf2796b2d9468d520e85f79ac834ffc2a18cadd17dff8ce00de23fc797d81",
   "contractKind": "non_executable_comfy_contract_resolution_ledger",
   "resolutions": [
     {
@@ -16499,14 +16499,14 @@
       "sha256": "sha256:d743007946b424026f86f0890bf7926953ad4808512b91018d100b4ea2940c75"
     },
     "templateAuthoringSpecs": {
-      "contentHash": "sha256:f7b2e5cdf11ffe88f38d298357d48641275c69d240ff13cf6b3d39eee3cdc522",
+      "contentHash": "sha256:25433f08dc2d43521b16c1b6ab53e9d0dfaeb78a1e794fee7c1695df65fa0dea",
       "path": "data/template-authoring-specs.v1.json",
-      "sha256": "sha256:b4fa65929304f22cba6f6d13bb138f7188d49115f81a859b29b9b603614d2f4b"
+      "sha256": "sha256:17ff9c9bbe8b7283c8ff960699d4f1cbe6baae15aaa8e487328ded4601359321"
     },
     "upstreamCoverage": {
-      "contentHash": "sha256:a7d36b205241f21d4067de7cc3b877c2c29538d3e2a2d965c03d670b9135f9ca",
+      "contentHash": "sha256:ed25c084c7fb0be603cc00e79e49d3b6dc53b053c88cd2bcd97711811eb239d1",
       "path": "data/upstream-coverage.v1.json",
-      "sha256": "sha256:6f5d0b3551c5c4c8e4f109947c8527c23f9c04d7d595ed134e2d8c82a6005cf0"
+      "sha256": "sha256:bfd3e3c6a565549bea2b7d545f78b89eda39129b14213f4323e73c94c8acfb83"
     },
     "workflowManifest": {
       "path": "data/workflow-library-manifest.json",

data/research/comfy-evidence-resolution.v1.json

@@ -16,7 +16,7 @@
       "presentationMediaHints"
     ]
   },
-  "contentHash": "sha256:f7fc575517953e6a7c0296851e9dde211de498a92aca36696867a50512f0634c",
+  "contentHash": "sha256:f18ffca9fe31ed2690729c1dad6cf4d82959a4f9b45c02b2336467c078cdc9a8",
   "contractKind": "non_executable_comfy_evidence_resolution_ledger",
   "resolutions": [
     {
@@ -7483,14 +7483,14 @@
       "sha256": "sha256:d743007946b424026f86f0890bf7926953ad4808512b91018d100b4ea2940c75"
     },
     "templateAuthoringSpecs": {
-      "contentHash": "sha256:f7b2e5cdf11ffe88f38d298357d48641275c69d240ff13cf6b3d39eee3cdc522",
+      "contentHash": "sha256:25433f08dc2d43521b16c1b6ab53e9d0dfaeb78a1e794fee7c1695df65fa0dea",
       "path": "data/template-authoring-specs.v1.json",
-      "sha256": "sha256:b4fa65929304f22cba6f6d13bb138f7188d49115f81a859b29b9b603614d2f4b"
+      "sha256": "sha256:17ff9c9bbe8b7283c8ff960699d4f1cbe6baae15aaa8e487328ded4601359321"
     },
     "upstreamCoverage": {
-      "contentHash": "sha256:a7d36b205241f21d4067de7cc3b877c2c29538d3e2a2d965c03d670b9135f9ca",
+      "contentHash": "sha256:ed25c084c7fb0be603cc00e79e49d3b6dc53b053c88cd2bcd97711811eb239d1",
       "path": "data/upstream-coverage.v1.json",
-      "sha256": "sha256:6f5d0b3551c5c4c8e4f109947c8527c23f9c04d7d595ed134e2d8c82a6005cf0"
+      "sha256": "sha256:bfd3e3c6a565549bea2b7d545f78b89eda39129b14213f4323e73c94c8acfb83"
     },
     "workflowManifest": {
       "path": "data/workflow-library-manifest.json",

data/template-authoring-specs.v1.json

@@ -14,14 +14,14 @@
     "selectsInputAssets": false,
     "studioVisible": false
   },
-  "contentHash": "sha256:f7b2e5cdf11ffe88f38d298357d48641275c69d240ff13cf6b3d39eee3cdc522",
+  "contentHash": "sha256:25433f08dc2d43521b16c1b6ab53e9d0dfaeb78a1e794fee7c1695df65fa0dea",
   "contractKind": "non_public_template_authoring_spec_ledger",
   "schemaVersion": 1,
   "sources": {
     "candidateContracts": {
-      "contentHash": "sha256:419ecd358329d4ee720c319bb634598870f8cd463fd3f832f0f23ff1b7818679",
+      "contentHash": "sha256:27df3a3e7b79a359a8333700983f334398a6317ee0427784204f5edc5682f331",
       "path": "data/template-candidate-contracts.v1.json",
-      "sha256": "sha256:61546d67c49ea8dfc089522fb222a6a024750fb11aea22df3274cbbf41da2e05"
+      "sha256": "sha256:144696f2c50db52b6b958d92b6b2fca4e366a4137d16f96dd29ec0d1abeb22ba"
     },
     "workflowManifest": {
       "path": "data/workflow-library-manifest.json",

data/template-candidate-contracts.v1.json

@@ -11,7 +11,7 @@
     "qualificationMetadataMeaning": "exact_manifest_binding_not_candidate_evidence",
     "studioVisible": false
   },
-  "contentHash": "sha256:419ecd358329d4ee720c319bb634598870f8cd463fd3f832f0f23ff1b7818679",
+  "contentHash": "sha256:27df3a3e7b79a359a8333700983f334398a6317ee0427784204f5edc5682f331",
   "contractKind": "non_public_template_candidate_contract_ledger",
   "contracts": [
     {
@@ -6687,9 +6687,9 @@
       "sha256": "sha256:f278e24ed314a1ab997b5521cc4c8b9576c67cdbbe76e19ffa3ce58a4593dc0e"
     },
     "upstreamCoverage": {
-      "contentHash": "sha256:a7d36b205241f21d4067de7cc3b877c2c29538d3e2a2d965c03d670b9135f9ca",
+      "contentHash": "sha256:ed25c084c7fb0be603cc00e79e49d3b6dc53b053c88cd2bcd97711811eb239d1",
       "path": "data/upstream-coverage.v1.json",
-      "sha256": "sha256:6f5d0b3551c5c4c8e4f109947c8527c23f9c04d7d595ed134e2d8c82a6005cf0"
+      "sha256": "sha256:bfd3e3c6a565549bea2b7d545f78b89eda39129b14213f4323e73c94c8acfb83"
     },
     "workflowManifest": {
       "path": "data/workflow-library-manifest.json",

data/upstream-coverage.v1.json

@@ -2339,7 +2339,7 @@
       "status": "executable"
     }
   ],
-  "contentHash": "sha256:a7d36b205241f21d4067de7cc3b877c2c29538d3e2a2d965c03d670b9135f9ca",
+  "contentHash": "sha256:ed25c084c7fb0be603cc00e79e49d3b6dc53b053c88cd2bcd97711811eb239d1",
   "diffusersPipelines": [
     {
       "artifactReviews": [],
@@ -8667,7 +8667,7 @@
     },
     "qualificationBoundary": "Executable means source/graph admission only; it does not imply live output, Auto, Gallery, hardware, license, or release qualification.",
     "templateBundle": "web/assets/studio-templates.js",
-    "templateBundleSha256": "e554ddc034cb069918936aae7641bf08345d86c5d3670494dee7e13739c570b8",
+    "templateBundleSha256": "015acbe27676173560cb135be77e5681f9cef766ab8f29fc0f0494840ad0275b",
     "transformers": {
       "productionRuntime": {
         "byteSize": 11625234,

docs/workflow-authoring-ux.md

@@ -27,6 +27,35 @@ conditioning or audio operations. Whole-pipeline routes expose Generate/Edit
 operations instead; those nodes do not claim independently replaceable denoising.
 Qwen-Image 2.1 uses its supported whole-pipeline route at the reviewed Diffusers pin.
 
+## Group nodes into a reusable Block
+
+Select nodes and choose **Create block**. New definitions use Block V2 and can
+contain existing Blocks such as Encode Inputs. Their nested contents, values and
+crossing connections are retained. Existing saved definitions are not rewritten.
+Creation errors appear inside the dialog. Public sockets and controls remain
+available; the dialog lets you choose controls and rename ports.
+
+Connection validation follows executable nodes inside containers. A loader and
+denoiser grouped around an external encoder can exchange values without creating
+an execution cycle. Required input checks also inspect these internal nodes; a
+missing model connection blocks Run and identifies the affected node or Block.
+
+When changing a model inside a Block, the change preview may retire untouched,
+automatically derived outputs that the new route cannot provide and nobody uses.
+Connected outputs, configured interfaces and preview bindings remain protected.
+Incompatible contracts require repair before the change can be applied.
+
+Scalar inputs declared with an editor (text, number, toggle or choices) retain
+both their editor and optional socket. A connection supplies the execution value
+and disables the literal editor; disconnecting restores the saved fallback.
+The source is shown when available. Model and media inputs retain their sockets.
+
+For a custom graph using Auto memory, **Graph ready · Auto check at Run** means
+static graph checks passed. Run checks the exact backend resource plan before
+submission. Use **Resources → Check Auto execution plan** to inspect it earlier.
+Installed model files alone do not qualify a recipe; Custom memory retains your
+chosen settings without claiming Auto qualification.
+
 ## Change the model
 
 ### Guidance control visibility

modiff/workflow_auto_resource.py

@@ -310,7 +310,15 @@ def _build_workflow_auto_plan(
                     "Use a covered batch size or select Expert."
                     if form.get("batchSize", 1) > 1 else None
                 )
-                raise ValueError(plan.get("blockingReason") or batch_reason or "No accepted recipe preserves this graph's model, precision and requested workload.")
+                reason = plan.get("blockingReason") or batch_reason or (
+                    "No accepted recipe preserves this graph's model, precision and requested workload. "
+                    f"Requested {repo}, {form['dtype']}, {form['quantizationMode']} quantization, "
+                    f"{form.get('width', 'default')}x{form.get('height', 'default')}, "
+                    f"{form.get('steps', 'default')} steps. "
+                    "Installed model files do not establish Auto resource qualification. "
+                    "Review the model's resource recipes or select Custom memory to keep your settings."
+                )
+                raise ValueError(reason)
             candidate = candidates[0]
             envelope = candidate.get("requirements", {})
             requirements = envelope

modules/ModularDiffusers/embeddings.py

@@ -86,6 +86,7 @@ class EncodePrompt(NodeBase):
     params = {
         "text_encoders": {
             "label": "Text Encoders *",
+            "required": True,
             "type": "diffusers_auto_models",
             "display": "input",
             "onSignal": "update_node",

tests/test_required_encoder_contract.py

@@ -0,0 +1,8 @@
+"""The registry must expose the model dependency before graph execution."""
+from modules.ModularDiffusers.embeddings import EncodePrompt
+
+
+def test_prompt_encoder_requires_models_but_accepts_inline_prompt():
+    assert EncodePrompt.params['text_encoders']['required'] is True
+    assert EncodePrompt.params['text_encoders']['display'] == 'input'
+    assert not EncodePrompt.params['prompt_input'].get('required', False)

tests/test_workflow_auto_resource.py

@@ -629,3 +629,18 @@ def test_owner_retention_normalizes_numeric_hardware_values(setup):
     result = plan(graph(two=True))
     assert result['canAutoRun']
     assert result['schedule'] is None
+
+
+def test_unqualified_recipe_explains_installed_files_are_not_auto_proof(setup):
+    plan, candidate, _, _ = setup
+    candidate['canAutoRun'] = False
+    g = graph()
+    before = deepcopy(g)
+    result = plan(g)
+    assert not result['canAutoRun']
+    message = ' '.join(result['issues'])
+    assert 'Installed model files do not establish Auto resource qualification' in message
+    assert 'test/model' in message
+    assert 'bfloat16' in message
+    assert 'Custom memory' in message
+    assert g == before