Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Blender Python add-on rules for operators, panels, properties, registration, testing, and API-safe scripting
.claude/skills/amariahak-blender-python-add-on-rules-for-operators-panels-properties-registration-tes/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 39% | 0% |
Blender Python add-on rules for operators, panels, properties, registration, testing, and API-safe scripting
Synced from https://github.com/PatrickJS/awesome-cursorrules/tree/main/rules/blender-python-addon.mdc.
__init__.py with clear register() and unregister() functions.bl_info or blender_manifest.toml according to the Blender version and packaging target.bpy.types.Operator for actions, bpy.types.Panel for UI, and bpy.types.PropertyGroup for grouped settings.bl_idname, bl_label, and bl_options explicitly.poll() before enabling operators.invoke() for interactive setup and execute() for the actual operation.{'FINISHED'} or {'CANCELLED'} consistently.PropertyGroup classes instead of loose global state.AddonPreferences.PointerProperty, CollectionProperty, and typed properties with names and descriptions.unregister().bmesh when editing mesh data programmatically.self.report() for user-facing operator feedback.draw() methods.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | fail→pass | 10,681 | 9,514 | -11% | 1 | 1 | 0% | 1,996 | 2,533 | +27% | 0 | 0 | — |
case-01 | fail→pass | 17,024 | 11,321 | -33% | 1 | 1 | 0% | 3,897 | 2,977 | -24% | 0 | 0 | — |
case-02 | pass→pass | 12,238 | 10,446 | -15% | 1 | 1 | 0% | 2,685 | 2,810 | +5% | 0 | 0 | — |
case-03 | pass→pass | 9,204 | 6,796 | -26% | 1 | 1 | 0% | 1,951 | 2,092 | +7% | 0 | 0 | — |
case-04 | pass→pass | 10,900 | 10,136 | -7% | 1 | 1 | 0% | 2,294 | 2,811 | +23% | 0 | 0 | — |
case-06 | pass→pass | 14,484 | 11,638 | -20% | 1 | 1 | 0% | 3,220 | 3,649 | +13% | 0 | 0 | — |
case-07 | pass→pass | 10,763 | 6,226 | -42% | 1 | 1 | 0% | 1,708 | 1,790 | +5% | 0 | 0 | — |
case-08 | fail→pass | 10,106 | 9,361 | -7% | 1 | 1 | 0% | 2,017 | 2,476 | +23% | 0 | 0 | — |
case-09 | fail→fail | 12,340 | 8,207 | -33% | 1 | 1 | 0% | 2,367 | 2,055 | -13% | 0 | 0 | — |
case-18 | pass→pass | 12,604 | 9,690 | -23% | 1 | 1 | 0% | 2,279 | 2,892 | +27% | 0 | 0 | — |
case-10 | pass→pass | 9,508 | 9,600 | +1% | 1 | 1 | 0% | 1,707 | 2,426 | +42% | 0 | 0 | — |
case-11 | pass→pass | 14,413 | 10,735 | -26% | 1 | 1 | 0% | 2,904 | 2,778 | -4% | 0 | 0 | — |
case-12 | fail→pass | 11,775 | 9,285 | -21% | 1 | 1 | 0% | 2,495 | 2,532 | +1% | 0 | 0 | — |
case-13 | fail→fail | 10,513 | 8,751 | -17% | 1 | 1 | 0% | 1,976 | 2,199 | +11% | 0 | 0 | — |
case-14 | pass→pass | 11,988 | 9,897 | -17% | 1 | 1 | 0% | 2,367 | 2,607 | +10% | 0 | 0 | — |
case-15 | fail→pass | 6,321 | 5,599 | -11% | 1 | 1 | 0% | 1,185 | 1,652 | +39% | 0 | 0 | — |
case-16 | fail→fail | 6,485 | 6,303 | -3% | 1 | 1 | 0% | 1,360 | 1,954 | +44% | 0 | 0 | — |
case-17 | pass→pass | 12,567 | 9,747 | -22% | 1 | 1 | 0% | 2,398 | 2,750 | +15% | 0 | 0 | — |
case-19 | pass→pass | 11,627 | 7,664 | -34% | 1 | 1 | 0% | 2,172 | 1,959 | -10% | 0 | 0 | — |
case-20 | pass→pass | 12,229 | 9,214 | -25% | 1 | 1 | 0% | 2,347 | 2,378 | +1% | 0 | 0 | — |
case-21 | pass→pass | 16,750 | 14,904 | -11% | 1 | 1 | 0% | 3,201 | 3,361 | +5% | 0 | 0 | — |
case-22 | fail→fail | 15,276 | 10,581 | -31% | 1 | 1 | 0% | 1,877 | 2,216 | +18% | 0 | 0 | — |
case-23 | fail→fail | 8,654 | 10,597 | +22% | 1 | 1 | 0% | 1,716 | 2,616 | +52% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +22 percentage points is the difference between those two pass rates over the 23 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.