chore: import upstream snapshot with attribution

showcase/integrations/claude-sdk-python/qa/tool-rendering-custom-catchall.md

@@ -0,0 +1,30 @@
+# QA: Tool Rendering (Custom Catch-all) — Claude Agent SDK (Python)
+
+## Prerequisites
+
+- Demo is deployed and accessible
+- Agent backend is healthy
+- ANTHROPIC_API_KEY is set
+
+## Test Steps
+
+### 1. Basic Functionality
+
+- [ ] Navigate to `/demos/tool-rendering-custom-catchall`
+- [ ] Verify the chat renders
+- [ ] Click "Weather in SF" — verify the agent calls `get_weather`
+- [ ] Verify the BRANDED `CustomCatchallRenderer` card appears (not the default)
+
+### 2. Feature-Specific Checks
+
+- [ ] Verify `data-testid="custom-catchall-card"` renders with the tool name
+- [ ] Verify status pill transitions: streaming → running → done
+- [ ] Try "Find flights" — verify the same branded card renders
+
+### 3. Error Handling
+
+- [ ] Verify no console errors
+
+## Expected Results
+
+- Every tool call renders via the custom `CustomCatchallRenderer`, including tool name, arguments pre, and result pre