docs: drop daslab.dev references from gallery + README

mirkokiefer · mirkokiefer · commit b3aebb22643c · 2026-05-12T17:58:18.000+07:00
scenecast is its own OSS project; the magnet-URL pitch belongs in the
benchmarks that point back here, not in scenecast's own surface.
diff --git a/README.md b/README.md
@@ -23,7 +23,7 @@ Ships with **7 canonical types** in core — abstract primitives any vendor can
 
 The **Mesh** type is the spatial seam — domain extensions like `protein/structure`, `robot/arm`, `printable/object`, `lab/labware` all `extends: ["mesh/file"]` and inherit format-discriminated rendering (`<model-viewer>` for glb/gltf/usdz today; Mol*, urdf-loader, three-stl-loader to follow). Same multi-target pipeline as Email — agents reason about a 3D asset in the same compact Markdown summary they use for everything else.
 
-**Vendor extensions ship with benchmarks, not core.** Gmail, Slack, Salesforce, SAP S/4HANA — these live in **benchmark-scoped repos** like [`scenebench`](https://github.com/daslabhq/scenebench), which delivers scenecast extensions for every vendor in its benchmark domain. Each benchmark gets a magnet URL on `daslab.dev` (e.g. `daslab.dev/labs/automationbench`, `daslab.dev/s4bench`) showing its tasks, vendor schemas, and a leaderboard — scrubbable and scored.
+**Vendor extensions ship with benchmarks, not core.** Gmail, Slack, Salesforce, SAP S/4HANA — these live in **benchmark-scoped repos** like [`scenebench`](https://github.com/daslabhq/scenebench), which delivers scenecast extensions for every vendor in its benchmark domain.
 
 Vendor types declare `extends: ["email/mailbox"]` etc. — tools that consume canonical types work uniformly across all vendors that implement them.
 
diff --git a/gallery/build.ts b/gallery/build.ts
@@ -755,7 +755,7 @@ const html = `<!DOCTYPE html>
       <h2>What you're looking at</h2>
       <p>Each card below is one <code>defineAsset()</code> declaration from the scenecast library. Every asset comes with a default <code>view</code> that renders state at <b>five sizes</b> (icon · small · medium · large · xlarge — modeled on Apple WidgetKit + Daslab's <code>WidgetSize</code>) and <b>five formats</b> (HTML · Markdown · Text · <a href="https://a2ui.org" target="_blank" rel="noreferrer noopener">A2UI v0.9</a> · <a href="https://blog.modelcontextprotocol.io/posts/2026-01-26-mcp-apps/" target="_blank" rel="noreferrer noopener">MCP Apps</a>).</p>
       <p>Same definition. Different consumers. Markdown is typically <b>3–5× cheaper in tokens</b> than dumping raw JSON; the icon view is ~95% cheaper. The two agent-UI standards that emerged in 2026 split the world: <b>A2UI</b> (Google, native-declarative — Lit/React/Flutter/OpenClaw) and <b>MCP Apps</b> (Anthropic + OpenAI, sandboxed iframe — Claude/ChatGPT/VS Code/Goose/Cursor). scenecast emits both. Every card shows our own HTML render, Google's reference <code>@a2ui/lit</code> renderer, and the same widget rendered inside a sandboxed MCP-Apps iframe — all consuming the exact same WidgetData JSON.</p>
-      <p>Vendor implementations (Gmail, Slack, Salesforce, SAP S/4HANA, …) ship from benchmark-scoped repos like <a href="https://github.com/daslabhq/scenebench">scenebench</a> as scenecast extensions for that benchmark's domain. Each benchmark owns its vendor types, scrubbable + scored at <a href="https://daslab.dev">daslab.dev</a>.</p>
+      <p>Vendor implementations (Gmail, Slack, Salesforce, SAP S/4HANA, …) ship from benchmark-scoped repos like <a href="https://github.com/daslabhq/scenebench">scenebench</a> as scenecast extensions for that benchmark's domain. Each benchmark owns its vendor types.</p>
       <div class="compat-row">
         <span class="label">A2UI camp</span>
         <a href="https://a2ui.org/reference/renderers/" target="_blank" rel="noreferrer noopener">@a2ui/lit</a>
diff --git a/gallery/index.html b/gallery/index.html
@@ -605,7 +605,7 @@
       <h2>What you're looking at</h2>
       <p>Each card below is one <code>defineAsset()</code> declaration from the scenecast library. Every asset comes with a default <code>view</code> that renders state at <b>five sizes</b> (icon · small · medium · large · xlarge — modeled on Apple WidgetKit + Daslab's <code>WidgetSize</code>) and <b>five formats</b> (HTML · Markdown · Text · <a href="https://a2ui.org" target="_blank" rel="noreferrer noopener">A2UI v0.9</a> · <a href="https://blog.modelcontextprotocol.io/posts/2026-01-26-mcp-apps/" target="_blank" rel="noreferrer noopener">MCP Apps</a>).</p>
       <p>Same definition. Different consumers. Markdown is typically <b>3–5× cheaper in tokens</b> than dumping raw JSON; the icon view is ~95% cheaper. The two agent-UI standards that emerged in 2026 split the world: <b>A2UI</b> (Google, native-declarative — Lit/React/Flutter/OpenClaw) and <b>MCP Apps</b> (Anthropic + OpenAI, sandboxed iframe — Claude/ChatGPT/VS Code/Goose/Cursor). scenecast emits both. Every card shows our own HTML render, Google's reference <code>@a2ui/lit</code> renderer, and the same widget rendered inside a sandboxed MCP-Apps iframe — all consuming the exact same WidgetData JSON.</p>
-      <p>Vendor implementations (Gmail, Slack, Salesforce, SAP S/4HANA, …) ship from benchmark-scoped repos like <a href="https://github.com/daslabhq/scenebench">scenebench</a> as scenecast extensions for that benchmark's domain. Each benchmark owns its vendor types, scrubbable + scored at <a href="https://daslab.dev">daslab.dev</a>.</p>
+      <p>Vendor implementations (Gmail, Slack, Salesforce, SAP S/4HANA, …) ship from benchmark-scoped repos like <a href="https://github.com/daslabhq/scenebench">scenebench</a> as scenecast extensions for that benchmark's domain. Each benchmark owns its vendor types.</p>
       <div class="compat-row">
         <span class="label">A2UI camp</span>
         <a href="https://a2ui.org/reference/renderers/" target="_blank" rel="noreferrer noopener">@a2ui/lit</a>
