feat: add animated GIFs + clean vault

GIFs (Remotion-rendered, FFmpeg palette-optimized):
- wiki/meta/wiki-graph-grow.gif (285KB, 6s, 15fps)
  Shows knowledge graph nodes appearing, coloring by type, edges connecting
  Color-coded: cyan=navigation, blue=concepts, purple=entities, green=sources
- wiki/meta/workflow-loop.gif (227KB, 5s, 15fps)
  Animated 4-step cycle: Drop Source → Claude Reads → Wiki Grows → Query
README: both GIFs shown side-by-side after the cover

Copyright clean-up (committed together):
- Removed Nate Herk source page and all 13 references
- Vault now 100% original content (3 concepts + 1 entity + navigation)

.obsidian/workspace.json

@@ -189,6 +189,9 @@
   },
   "active": "c0d470cf50ffdf79",
   "lastOpenFiles": [
+    "README.md.tmp.224546.1775555737152",
+    "wiki/meta/workflow-loop.gif",
+    "wiki/meta/wiki-graph-grow.gif",
     "wiki/Wiki Map.canvas.tmp.224546.1775555483542",
     "wiki/entities/Andrej Karpathy.md.tmp.224546.1775555435117",
     "wiki/concepts/Hot Cache.md.tmp.224546.1775555433665",
@@ -206,11 +209,8 @@
     "_attachments/images/skool-hub/sq-portrait-clean.png",
     "_attachments/images/skool-hub/sq-10.png",
     "_attachments/images/skool-hub/sq-01.png",
-    "_attachments/images/skool-hub/pro10-var-20.png",
-    "_attachments/images/skool-hub/pro10-var-10.png",
     "_attachments/images/skool-hub",
     "_attachments/images",
-    "_attachments",
     "wiki/entities/_index.md",
     "wiki/meta/dashboard.md",
     "wiki/sources/_index.md",

README.md

@@ -12,6 +12,11 @@ Based on Andrej Karpathy's LLM Wiki pattern.
 
 ## What It Does
 
+<p align="center">
+  <img src="wiki/meta/wiki-graph-grow.gif" alt="Knowledge graph growing" width="48%" />
+  <img src="wiki/meta/workflow-loop.gif" alt="Workflow loop" width="48%" />
+</p>
+
 You drop sources. Claude reads them, extracts entities and concepts, updates cross-references, and files everything into a structured Obsidian vault. The wiki gets richer with every ingest.
 
 You ask questions. Claude reads the hot cache (recent context), scans the index, drills into relevant pages, and synthesizes an answer. It cites specific wiki pages, not training data.