@deepblend/dsh-blender-preset 0.9.0 → 0.11.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@deepblend/dsh-blender-preset",
|
|
3
|
-
"version": "0.
|
|
3
|
+
"version": "0.11.0",
|
|
4
4
|
"type": "module",
|
|
5
5
|
"description": "DeepBlend Agent-preset plane: ships the DeepBlend Studio agent presets and the deployer row that installs them into the DSH home.",
|
|
6
6
|
"main": "lib/index.js",
|
|
@@ -86,6 +86,13 @@
|
|
|
86
86
|
a SceneSpec document, every accepted change is an immutable revision, and you
|
|
87
87
|
look at a scene with previews before you commit hours to a final render.
|
|
88
88
|
Load the deepblend-studio skill before starting real work.
|
|
89
|
+
For an existing dimension edit, finish with only a compact factual receipt:
|
|
90
|
+
saved project/revisions, requested native overall spans, protected timing/appearance,
|
|
91
|
+
registered files, and remaining score findings or unmeasured conditions.
|
|
92
|
+
Do not add component tables or a retelling of movement groups. Do not claim all
|
|
93
|
+
part sizes stayed unchanged when generatorChangedEntityIds is nonempty.
|
|
94
|
+
Beauty images do not certify no floating, no interpenetration, topology or clearance;
|
|
95
|
+
do not add those claims to the handoff without explicit corresponding measurements.
|
|
89
96
|
suffix: Your working directory is {{cwd}}.
|
|
90
97
|
|
|
91
98
|
- id: agent-instructions
|
|
@@ -54,6 +54,24 @@ checkpoint and preview together. When a result says its image was not attached,
|
|
|
54
54
|
the model has not seen those pixels; retain the saved artifact and report that
|
|
55
55
|
limit. Additional views still need to show the features required by the brief.
|
|
56
56
|
|
|
57
|
+
For an existing assembly dimension edit, the final handoff must be a compact receipt:
|
|
58
|
+
- project/current revision and the retained original;
|
|
59
|
+
- only the requested overall dimensions, from the saved native span receipt;
|
|
60
|
+
- saved custom keyframe timing and protected cameras when requested;
|
|
61
|
+
- the actual registered native file and primary/back images;
|
|
62
|
+
- the remaining score findings and unmeasured global conditions.
|
|
63
|
+
Do not add component tables, inferred margins, or a prose retelling of movement groups.
|
|
64
|
+
Do not infer no floating, no interpenetration, topology or clearance from beauty previews.
|
|
65
|
+
Without corresponding measurements those conditions remain unassessed, even if a view looks good.
|
|
66
|
+
The save receipt and scene document contain the detailed part changes for inspection.
|
|
67
|
+
The save receipt's generatorChangedEntityIds explicitly lists structurally resized parts.
|
|
68
|
+
When it is nonempty, never say "no part size changed" or "all part geometry unchanged".
|
|
69
|
+
Preservation applies to the original revision and independent hand edits, not to the
|
|
70
|
+
structural generator dimensions requested to change.
|
|
71
|
+
Structural shells, openings, baffles and panels grow: the recipe's fixed-part wording is
|
|
72
|
+
not a claim that those structural dimensions stay unchanged. If the user explicitly asks
|
|
73
|
+
for a detailed component report, quote actual receipt fields individually and retain gaps.
|
|
74
|
+
|
|
57
75
|
For a later dimensional change, read `blender_recipe_parameters`, then save
|
|
58
76
|
only the changed parameter IDs with the current base revision and preview
|
|
59
77
|
enabled. This preserves independent appearance, the original revision and
|
|
@@ -86,9 +104,57 @@ from actual surface measurements. Without a separate supported surface check,
|
|
|
86
104
|
report wall thickness and clearance as unmeasured; do not give a numeric gap or
|
|
87
105
|
claim "basically preserved" from bounding dimensions or preview pixels.
|
|
88
106
|
|
|
107
|
+
For actual local surface evidence, the same tool optionally accepts
|
|
108
|
+
`geometrySurfaceSamples` (1–8). Each sample supplies `sampleId`, a selected
|
|
109
|
+
`entityId`, an outside `originMeters`, a non-zero `direction`, and explicit
|
|
110
|
+
`searchDistanceMm` (0.01–500); optional `targetEntityId` must be another selected
|
|
111
|
+
part. Choose seeds from actual saved bounds and the requested region, not from
|
|
112
|
+
declared thickness or gap values. The runtime follows the first entry face's
|
|
113
|
+
normal to its first exit and, if requested, the target's first entry. Keep seeds,
|
|
114
|
+
frame, part IDs and source SHA with the receipt; use the same explicit seeds to
|
|
115
|
+
compare saved revisions. A failed sample is unavailable, never zero.
|
|
116
|
+
|
|
117
|
+
Use `geometryOnly:true` with the native selection for a compact measurement receipt;
|
|
118
|
+
it keeps the complete verified objects, samples and source identity while omitting
|
|
119
|
+
the scene catalogue. Read the scene separately when selecting parts or seeds.
|
|
120
|
+
It cannot be combined with `full:true`.
|
|
121
|
+
|
|
122
|
+
For before/after comparisons, supply `geometryCompareRevision` with the old saved
|
|
123
|
+
revision and the same explicit samples on the current read. The tool verifies both
|
|
124
|
+
sources and computes `geometryComparison.samples` changes (current minus reference)
|
|
125
|
+
and maxima over **all** comparable requested samples. Chord-only samples still count
|
|
126
|
+
in the chord maximum. Unavailable pairs are omitted from numeric changes, never zero.
|
|
127
|
+
Use these computed summaries rather than mental arithmetic. Preserve sample IDs,
|
|
128
|
+
coverage counts and unavailable rows; round only for display. Do not assert left/right
|
|
129
|
+
symmetry or an explanation for a change from this summary. A world seed does not
|
|
130
|
+
necessarily hit the same material point across revisions. If the comparison is
|
|
131
|
+
unavailable, report the raw evidence and limit the claim instead of guessing maxima.
|
|
132
|
+
The comparison also computes percent changes against each reference value. Quote
|
|
133
|
+
the separate chord and separation percentage maxima exactly (round only for display).
|
|
134
|
+
Do not apply a chord percentage to the gap, derive percentages from rounded table
|
|
135
|
+
values, or introduce a combined relative bound. If no percentage receipt exists,
|
|
136
|
+
omit the percentage claim. Do not equate a measured change with user approval of
|
|
137
|
+
its size; state the observed values and coverage, and leave unmeasured conditions open.
|
|
138
|
+
|
|
139
|
+
`normalSurfaceChordMm` runs from source entry to source exit. Target
|
|
140
|
+
`normalRaySeparationMm` runs from source **EXIT** to target entry along that
|
|
141
|
+
inward normal, excluding the source chord. Describe those two surfaces explicitly;
|
|
142
|
+
do not label it as source outer surface to target entry or as global free clearance.
|
|
143
|
+
Triangle-face normals differ from shading normals. These samples do not prove
|
|
144
|
+
global minimum wall thickness, whole-assembly free clearance, no collision,
|
|
145
|
+
manufacturing safety or visual quality. Other scene objects and mesh
|
|
146
|
+
self-intersections are not certified. Chord/target searches each stop at 100 mm;
|
|
147
|
+
unsupported topology, overlap, occlusion and missed rays must remain explicit.
|
|
148
|
+
If no supported samples cover a requested condition, keep it unmeasured.
|
|
149
|
+
|
|
89
150
|
Inspect the saved preview needed for the brief; add a targeted view if it does
|
|
90
151
|
not show the requested features. Once the requested native file and preview
|
|
91
152
|
are saved, conclude with the project, revision and actual delivery paths.
|
|
153
|
+
Use the current operation's returned project-relative paths and identify the project
|
|
154
|
+
and revision. A historical producer path does not establish the current store root.
|
|
155
|
+
Do not guess absolute paths from old metadata. If a generic presentation tool cannot
|
|
156
|
+
register a project-relative file, report its path and use the saved workbench preview;
|
|
157
|
+
do not invent a successful attachment or search unrelated stores to fabricate one.
|
|
92
158
|
Separate verified dimensions/topology from appearance observations. An
|
|
93
159
|
unavailable vision reviewer is a limitation to report, not a reason to repeat
|
|
94
160
|
unchanged renders or to claim artistic approval. Additional artistic review
|
|
@@ -230,8 +296,22 @@ the current revision and repeat the affected views with the same settings.
|
|
|
230
296
|
the first revision is `r0001`. If no recipe fits, use a supplied scene or the minimal scaffold route.
|
|
231
297
|
For an existing assembly, use **`blender_recipe_parameters`** with `action:read` first.
|
|
232
298
|
Use the returned revision, numeric parameter IDs and allowed ranges in `action:save`.
|
|
299
|
+
Read/save receipts include linkedAnimationTracks from the selected saved SceneSpec. Use its
|
|
300
|
+
actual keyframe frames and values to verify custom timing and travel; recipe defaults, shot
|
|
301
|
+
prose and the observation/render frame are not the current animation keyframe list.
|
|
233
302
|
This creates one revision in the same project, using ordinary patches internally; it preserves
|
|
234
303
|
independent materials, bevels, photography and history. Do not recreate the project to resize it.
|
|
304
|
+
Save receipts include placementChanges: groups of actual saved object-origin deltas in mm,
|
|
305
|
+
with entity IDs and from/to revision. Quote the group containing each named part; front
|
|
306
|
+
controls and rear connectors may move in opposite directions. These SceneSpec deltas
|
|
307
|
+
are not native surface distances or clearances.
|
|
308
|
+
A part following an assembly edge can change its offset from the cabinet centre. Do not
|
|
309
|
+
describe all relative offsets as invariant. A world-coordinate endpoint is not an exposure
|
|
310
|
+
or clearance distance. If no returned field defines a relationship measurement, leave it
|
|
311
|
+
unmeasured instead of deriving an attachment distance or margin in the final answer.
|
|
312
|
+
Pinned speaker 2.0/2.1 assemblies can retain a complete manual translation, including the
|
|
313
|
+
button rest anchor, while resizing. Partial moves and rotated dimension-controlled parts
|
|
314
|
+
remain conflicts. Keep custom photography; verify its framing after the size change.
|
|
235
315
|
Controlled geometry conflicts require resolving the named parts before saving. Old recipes
|
|
236
316
|
without geometry controls return available:false. Use a fresh idempotencyKey for each new
|
|
237
317
|
intent, including edits after restoring history; reuse it only for the same retry.
|
|
@@ -239,6 +319,9 @@ the current revision and repeat the affected views with the same settings.
|
|
|
239
319
|
Declare its returned asset descriptor in a patch, then bind its intended use:
|
|
240
320
|
a model instance, material map, environment, or review reference. See the model
|
|
241
321
|
workflow below and the reference image guide above.
|
|
322
|
+
`blender_project_get.projectDirectory` is the active project directory, including after
|
|
323
|
+
relocation or restore. Combine it with returned project-relative artifact paths for file
|
|
324
|
+
registration. Do not guess a producer's old absolute directory from historical metadata.
|
|
242
325
|
4. **`blender_scene_get`** — read before writing. Patching without reading is how a
|
|
243
326
|
stale `baseRevision` produces a refusal you then have to diagnose.
|
|
244
327
|
5. **`blender_scene_patch`** — submit ordinary scene edits as bounded operations. There is no free-form
|
|
@@ -318,6 +401,10 @@ hard to get wrong:
|
|
|
318
401
|
- **The score and the issues are measured, not reported.** The host computes them
|
|
319
402
|
deterministically from the rendered pixels — composition, exposure, occlusion. The
|
|
320
403
|
same scene scores the same twice, and a contract test asserts that.
|
|
404
|
+
Keep measured open issues visible in the handover. An unexpected threshold or
|
|
405
|
+
plausible explanation does not establish a false positive. Distinguish a possible
|
|
406
|
+
limitation from a verified cause; dismiss an issue only with separate evidence
|
|
407
|
+
from the relevant view and object.
|
|
321
408
|
- **You are the only one who can see meaning.** The measurements cannot tell a watch
|
|
322
409
|
dial that is *supposed* to be bright from one that is blown out, or a screen that is
|
|
323
410
|
*supposed* to be hidden from one that is occluded by mistake. When a scene measures
|