fix: review polish for the four newest examples + documentation currency (#51)

Example fixes from review of the auto-generated batch:
- gn-instance-grid and temp-override-join reframed — the right cube
  column and the top step were clipped at the frame edge; both now sit
  fully in frame with margin (heroes and previews re-rendered)
- temp-override-join drops the redundant select_set/active-object setup:
  the temp_override alone fabricates the operator context, which is the
  contract the example teaches
- curve-bevel-arc stops calling its vert/face constants closed-form:
  they are labeled as a measured regression gate with re-measure
  instructions (curve tessellation has no simple formula)

Documentation currency (examples existed nowhere in the core docs):
- CLAUDE.md: examples/scripts/docs-gallery in the architecture block, a
  new Examples (12) section with the authoring/wiring contract, examples
  in release-hygiene step 1, site-generation note in the workflow section
- AGENTS.md: content-base list de-pinned from 'as of v0.2.0' and gains
  the examples bullet, structure block gains examples/scripts/docs
  entries, blender-smoke.yml documented in CI/CD
- README: header, badge row, and overview now count 12 examples
  (validate-counts language preserved)
- ROADMAP: the fleet-template lift-and-shift candidate marked resolved
  differently — vendored locally, gallery.json is permanent

Both touched examples re-verified headless on Blender 5.1.1 and 4.4.3.

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

Author: TMHSDigital

AGENTS.md

@@ -6,14 +6,15 @@ Guidance for AI coding agents working on the Blender Developer Tools repository.
 
 ## Repository overview
 
-Skills, rules, snippets, and a starter template for Blender Python development.
+Skills, rules, snippets, starter templates, and runnable smoke-gated examples
+for Blender Python development.
 The repo targets **Blender 5.1** (current stable) with a **Blender 4.5 LTS**
 fallback. There is no MCP server. It ships a `.cursor-plugin/plugin.json`
 manifest so the ecosystem drift checker classifies it as a `cursor-plugin`.
 This is content the AI loads when the user asks Blender questions or works on
 Blender add-ons in Cursor or Claude Code.
 
-The content base as of v0.2.0:
+The content base (counts are CI-enforced against README.md and the manifest):
 
 - 12 skills covering scaffolding, operators, panels, properties, mesh and
   bmesh, headless batch scripts, slotted-actions animation (5.x), programmatic
@@ -26,6 +27,11 @@ The content base as of v0.2.0:
 - 2 templates: `extension-addon-template` for Extensions Platform add-ons,
   and `headless-batch-script-template` for unattended batch jobs.
 - 17 snippets covering canonical patterns.
+- 12 examples under `examples/<name>/`: runnable scripts that assert a real
+  API contract with deterministic checks, exit non-zero on failure, and
+  optionally render a still via `--output`. Each is executed headless on
+  Blender 4.5 LTS and 5.1 by `blender-smoke.yml`; its render ships in the
+  site gallery. Anatomy and authoring rules: copy `examples/bmesh-gear/`.
 
 ## Repository structure
 
@@ -35,7 +41,11 @@ Blender-Developer-Tools/
   rules/<rule-name>.mdc          # 6 rule files
   templates/<template-name>/     # 2 starter templates
   snippets/<snippet-name>.py     # 17 standalone Python snippets
-  .github/workflows/             # validate, drift-check, release, label-sync
+  examples/<name>/               # 12 runnable smoke-gated examples (+ gallery.json)
+  scripts/build_gallery.py       # generates docs/gallery/ (stdlib only)
+  scripts/site/                  # vendored landing-page build (build_site.py + template)
+  docs/gallery/                  # committed generated gallery pages + hero assets
+  .github/workflows/             # validate, blender-smoke, drift-check, release, pages, label-sync
   .github/dependabot.yml
   AGENTS.md, CLAUDE.md, README.md, ROADMAP.md, CHANGELOG.md
   CONTRIBUTING.md, SECURITY.md, CODE_OF_CONDUCT.md
@@ -116,6 +126,10 @@ way, and a one-paragraph rationale. 30 to 80 lines is the right size.
   every skill, rule, snippet, template, and example on disk must be listed,
   and the manifest `version` must equal `VERSION`. The release pipeline owns
   the manifest `version` line (see `release.yml` below) — never hand-edit it.
+- `blender-smoke.yml` executes every shipped example (check-only, no render)
+  plus snippet/template smoke tests inside REAL headless Blender, on both
+  4.5 LTS and 5.1, on every PR and a weekly schedule. A new example is not
+  shipped until it has a step here.
 - `drift-check.yml` consumes `Developer-Tools-Directory/.github/actions/
   drift-check@v1.15` to enforce ecosystem standards-version markers.
 - `release.yml` auto-bumps the version, tags, force-updates floating tags

CLAUDE.md

@@ -6,7 +6,7 @@ This file provides guidance to Claude Code (claude.ai/code) when working with co
 
 ## Project Overview
 
-The **Blender Developer Tools** repository is at **v0.9.1**. It packages skills, rules, snippets, and starter templates for Blender Python development with Cursor and Claude Code. Coverage targets **Blender 5.1** (current stable) with **Blender 4.5 LTS** fallback. There is no MCP server; content is consumed directly by the AI when working in Blender add-on or scripting projects.
+The **Blender Developer Tools** repository is at **v0.9.1**. It packages skills, rules, snippets, starter templates, and runnable smoke-gated examples for Blender Python development with Cursor and Claude Code. Coverage targets **Blender 5.1** (current stable) with **Blender 4.5 LTS** fallback. There is no MCP server; content is consumed directly by the AI when working in Blender add-on or scripting projects.
 
 **Version:** 0.9.1
 **License:** CC-BY-NC-ND-4.0
@@ -19,6 +19,10 @@ skills/<skill-name>/SKILL.md   - AI workflow definitions, 12 total
 rules/<rule-name>.mdc          - Anti-pattern rules, 6 total
 templates/<template-name>/     - Starter projects, 2 total
 snippets/<snippet-name>.py     - Standalone code patterns, 17 total
+examples/<name>/               - Runnable smoke-gated examples, 12 total (+ gallery.json)
+scripts/build_gallery.py       - Regenerates docs/gallery/ from gallery.json (stdlib only)
+scripts/site/                  - Vendored landing-page build (Jinja2)
+docs/gallery/                  - Committed generated gallery pages + hero renders
 VERSION                        - Source of truth for the repo version
 ```
 
@@ -76,9 +80,24 @@ v0.1.0: canonical object creation and deletion, depsgraph evaluated mesh, bmesh
 
 v0.2.0: Principled BSDF material, driver-with-custom-function via `driver_namespace`, application handler registration, shader node group with cross-version `interface` API, `foreach_get` bulk vertex read, version-branch skeleton, and USD export with `evaluation_mode='RENDER'`.
 
+## Examples (12)
+
+Runnable scripts at `examples/<name>/`, each asserting a real API contract with
+deterministic checks (exit non-zero on failure) and optionally rendering a still via
+`--output`. All twelve run headless on Blender 4.5 LTS and 5.1 in `blender-smoke.yml`;
+their renders ship in the site gallery at `docs/gallery/`. `examples/gallery.json` is the
+gallery's source of truth. When authoring a new one, copy the anatomy of
+`examples/bmesh-gear/` (script structure, README shape, dark-studio render recipe) and
+wire all of: gallery.json entry, `.cursor-plugin/plugin.json` examples array (CI-gated),
+a `blender-smoke.yml` step, a README gallery row, hero webp (1280×720) in
+`docs/gallery/assets/` + preview webp (1200×675), then run `python scripts/build_gallery.py`.
+
 ## Development Workflow
 
-This is a content repository, no build step. Edit `SKILL.md`, `.mdc`, `.py`, and `.toml` files directly.
+This is a content repository, no build step for skills/rules/snippets/templates — edit
+`SKILL.md`, `.mdc`, `.py`, and `.toml` files directly. The website is generated:
+`scripts/build_gallery.py` (stdlib) regenerates `docs/gallery/` and must be re-run after
+touching `examples/`; the landing page builds from `scripts/site/` at deploy time.
 
 The AI consumes content via:
 
@@ -109,7 +128,7 @@ The release pipeline is automated via `release.yml` on push to `main` for conten
 
 When adding content to a future version:
 
-1. Add files under `skills/`, `rules/`, `snippets/`, or `templates/`.
+1. Add files under `skills/`, `rules/`, `snippets/`, `templates/`, or `examples/`.
 2. Update README.md aggregate counts (the `validate-counts` job enforces correctness).
 3. Update ROADMAP.md candidate pool entries.
 4. Use `feat:` for new content, `fix:` for corrections.

README.md

@@ -3,7 +3,7 @@
 ---
 
 <p align="center">
-  <strong>Skills, rules, snippets, and a starter template for Blender Python development</strong>
+  <strong>Skills, rules, snippets, templates, and runnable examples for Blender Python development</strong>
 </p>
 
 <p align="center">
@@ -17,14 +17,14 @@
 </p>
 
 <p align="center">
-  <strong>12 skills</strong> &nbsp;&bull;&nbsp; <strong>6 rules</strong> &nbsp;&bull;&nbsp; <strong>2 templates</strong> &nbsp;&bull;&nbsp; <strong>17 snippets</strong>
+  <strong>12 skills</strong> &nbsp;&bull;&nbsp; <strong>6 rules</strong> &nbsp;&bull;&nbsp; <strong>2 templates</strong> &nbsp;&bull;&nbsp; <strong>17 snippets</strong> &nbsp;&bull;&nbsp; <strong>12 examples</strong>
 </p>
 
 ---
 
 ## Overview
 
-This repository ships **12 skills, 6 rules, 2 templates, and 17 snippets** for Blender Python development targeting Blender 5.1 (current stable) with Blender 4.5 LTS fallback support.
+This repository ships **12 skills, 6 rules, 2 templates, 17 snippets, and 12 runnable examples** for Blender Python development targeting Blender 5.1 (current stable) with Blender 4.5 LTS fallback support.
 
 The content is consumed by AI coding agents (Cursor, Claude Code, any MCP-capable client) when working on Blender add-ons, geometry nodes scripts, batch pipelines, or animation tooling. There is no build step. Edit the markdown and Python files directly.
 

ROADMAP.md

@@ -89,7 +89,7 @@ Audit pass on v0.1.0 content: standards-version markers bumped from `1.9.1` to `
 
 Not committed; target list for the next content version. (v0.3.0 shipped the smoke-gated `examples/` track.)
 
-- **Fleet Pages facelift + examples support** (infra) -- ship as ONE coordinated change to the meta-repo `site-template/template.html.j2` + `build_site.py`, since both need a template edit and two fleet-wide pushes is worse than one. (a) examples discovery: load `examples/gallery.json` and render an Examples grid + a nav link to it (the nav link closes the landing->gallery cross-link gap, impossible today without a template edit); (b) landing facelift adopting the direction proven by this repo's gallery (shared light/dark tokens, fluid hero type scale, the card system, `:focus-visible`, reduced-motion). This repo's local gallery (`examples/gallery.json` + `docs/gallery/`, see `docs/gallery/DESIGN_NOTES.md`) is the **prototype and convergence target**: once the shared template reads the same `gallery.json` schema, the local generator (`scripts/build_gallery.py`) and page are retired -- a lift-and-shift, not a rewrite. The Option-2 cycle must read the full template-consumer set, account for floating-main consumption (every repo updates on next deploy), and prove backward compatibility (a repo with no `gallery.json` renders unchanged) before any meta-repo merge. No confirmed rendering bug today (the suspected Skills/Rules overlap was verified to be a normal collapsed accordion), so the cross-link discoverability gap is the main driver.
+- ~~Fleet Pages facelift + examples support~~ **RESOLVED differently (2026-07-03)**: the meta-repo migration was dropped — the fleet template only scaffolds new tools, and each tool's site evolves independently after that. This repo vendored the site build into `scripts/site/`, redesigned landing + gallery as the Blender-viewport system (see `docs/gallery/DESIGN_NOTES.md`), added the examples grid, nav link, and full hero stats locally. `scripts/build_gallery.py` and `examples/gallery.json` are now permanent, not a prototype awaiting lift-and-shift.
 - `modal-operators` skill -- `invoke` returning `RUNNING_MODAL`, the `modal()` event handler, modal cancellation patterns
 - `usd-pipelines` skill -- USD export options, `evaluation_mode`, instancing, the USD vs glTF tradeoffs
 - `mathutils-patterns` skill -- `mathutils.Vector`, `Matrix`, `Quaternion`, common transforms, the `@` operator

docs/gallery/assets/gn-instance-grid-hero.webp

@@ -205,9 +205,11 @@ <h2>Source</h2>
 Witnesses that renderable tubes are authored on `bpy.types.Curve` directly
 (`splines.new(&#x27;BEZIER&#x27;)`, `bezier_points`, `bevel_depth`, `use_fill_caps`) —
 not by meshing first or calling curve operators. The check asserts the
-closed-form point count, bevel depth, filled-cap topology, and that the
-depsgraph-evaluated mesh has a Z span of a tube whose centerline sits at
-`z = bevel_depth` (resting on the floor).
+closed-form point count and bevel depth, the closed-form Z span (tube
+centerline at `z = bevel_depth`, resting on the floor) and X span, plus the
+evaluated vert/face counts as a MEASURED regression gate — curve tessellation
+has no simple closed form, so those two constants pin today&#x27;s behavior (see
+EXPECT_VERTS below for how to re-measure if a future Blender retessellates).
 
 By default it runs only the correctness check (no render) — the CI smoke
 check. Pass --output to also render a still:
@@ -222,7 +224,10 @@ <h2>Source</h2>
 BEVEL = <span class="n">0.15</span>
 BEVEL_RES = <span class="n">4</span>
 RES_U = <span class="n">12</span>
-<span class="c"># measured for the parameters above with use_fill_caps=True — identical on 4.4 and 5.1</span>
+<span class="c"># MEASURED regression constants, not closed-form: curve-to-mesh tessellation</span>
+<span class="c"># (rings x bevel segments + cap fans) has no simple formula. Verified identical</span>
+<span class="c"># on 4.4, 4.5 LTS, and 5.1. If a future Blender changes tessellation, re-measure</span>
+<span class="c"># by printing len(em.vertices)/len(em.polygons) in check() and update these.</span>
 EXPECT_VERTS = <span class="n">1044</span>
 EXPECT_FACES = <span class="n">1028</span>
 

docs/gallery/assets/temp-override-join-hero.webp

@@ -405,9 +405,9 @@ <h2>Source</h2>
     light(<span class="s">&quot;Rim&quot;</span>, (<span class="n">1.5</span>, <span class="n">4.5</span>, <span class="n">2.0</span>), <span class="n">480.0</span>, <span class="n">4.0</span>, (<span class="n">1.0</span>, <span class="n">0.75</span>, <span class="n">0.45</span>))
 
     cam_data = bpy.data.cameras.new(<span class="s">&quot;Cam&quot;</span>)
-    cam_data.lens = <span class="n">55.0</span>
+    cam_data.lens = <span class="n">50.0</span>
     cam = bpy.data.objects.new(<span class="s">&quot;Cam&quot;</span>, cam_data)
-    cam.location = (<span class="n">3.6</span>, -<span class="n">4.2</span>, <span class="n">2.8</span>)
+    cam.location = (<span class="n">4.6</span>, -<span class="n">5.4</span>, <span class="n">3.4</span>)
     scene.collection.objects.link(cam)
     scene.camera = cam
     track = cam.constraints.new(<span class="s">&#x27;TRACK_TO&#x27;</span>)

docs/gallery/curve-bevel-arc/index.html

@@ -250,15 +250,11 @@ <h2>Source</h2>
 
 
 <span class="k">def</span> join_with_temp_override(target, sources):
-    <span class="s">&quot;&quot;&quot;The contract this example witnesses: temp_override, not context.copy().&quot;&quot;&quot;</span>
-    view_layer = bpy.context.view_layer
-    <span class="k">for</span> obj <span class="k">in</span> bpy.data.objects:
-        obj.select_set(<span class="k">False</span>)
-    target.select_set(<span class="k">True</span>)
-    <span class="k">for</span> src <span class="k">in</span> sources:
-        src.select_set(<span class="k">True</span>)
-    view_layer.objects.active = target
+    <span class="s">&quot;&quot;&quot;The contract this example witnesses: temp_override, not context.copy().
 
+    The override alone fabricates the whole operator context — no select_set,
+    no view_layer.objects.active. That is the point: the scene&#x27;s real
+    selection state stays untouched.&quot;&quot;&quot;</span>
     <span class="k">with</span> bpy.context.temp_override(
         active_object=target,
         selected_objects=[target, *sources],
@@ -374,7 +370,7 @@ <h2>Source</h2>
     cam_data = bpy.data.cameras.new(<span class="s">&quot;Cam&quot;</span>)
     cam_data.lens = <span class="n">50.0</span>
     cam = bpy.data.objects.new(<span class="s">&quot;Cam&quot;</span>, cam_data)
-    cam.location = (<span class="n">4.2</span>, -<span class="n">5.2</span>, <span class="n">4.0</span>)
+    cam.location = (<span class="n">5.0</span>, -<span class="n">6.4</span>, <span class="n">4.6</span>)
     scene.collection.objects.link(cam)
     scene.camera = cam
     track = cam.constraints.new(<span class="s">&#x27;TRACK_TO&#x27;</span>)

docs/gallery/gn-instance-grid/index.html

@@ -3,9 +3,11 @@
 Witnesses that renderable tubes are authored on `bpy.types.Curve` directly
 (`splines.new('BEZIER')`, `bezier_points`, `bevel_depth`, `use_fill_caps`) —
 not by meshing first or calling curve operators. The check asserts the
-closed-form point count, bevel depth, filled-cap topology, and that the
-depsgraph-evaluated mesh has a Z span of a tube whose centerline sits at
-`z = bevel_depth` (resting on the floor).
+closed-form point count and bevel depth, the closed-form Z span (tube
+centerline at `z = bevel_depth`, resting on the floor) and X span, plus the
+evaluated vert/face counts as a MEASURED regression gate — curve tessellation
+has no simple closed form, so those two constants pin today's behavior (see
+EXPECT_VERTS below for how to re-measure if a future Blender retessellates).
 
 By default it runs only the correctness check (no render) — the CI smoke
 check. Pass --output to also render a still:
@@ -20,7 +22,10 @@
 BEVEL = 0.15
 BEVEL_RES = 4
 RES_U = 12
-# measured for the parameters above with use_fill_caps=True — identical on 4.4 and 5.1
+# MEASURED regression constants, not closed-form: curve-to-mesh tessellation
+# (rings x bevel segments + cap fans) has no simple formula. Verified identical
+# on 4.4, 4.5 LTS, and 5.1. If a future Blender changes tessellation, re-measure
+# by printing len(em.vertices)/len(em.polygons) in check() and update these.
 EXPECT_VERTS = 1044
 EXPECT_FACES = 1028