Brain map
See how your memories connect: supersede chains, co-access, shared entities and (optionally) semantic similarity.
Two views
| View | Where | Engine |
|---|---|---|
| Simplified map (2D) | /app?view=brain — the Brain tab | sigma.js and graphology, force layout in a worker |
| Full 3D Map | /app/brain | three.js with instanced rendering, layout in a worker |
What the edges mean
| Edge type | Meaning |
|---|---|
supersedes | A memory and the older version it replaced (superseded_by is the reverse) |
co_access | Memories that were recalled together |
shared_entity | Memories that mention the same entity |
semantic | Embedding similarity ≥ 0.7, top 3 per node; only when semantic=true is requested |
Render budget
The stats bar has Nodes and Edges budget selectors (defaults 5000 nodes and 20000 edges). Changing a budget does not re-run the layout, which always covers every node.
Data source
Both views read GET /api/v1/graph. Parameters: center (memory id to start from, default most recent), hops (default 3, max 5), limit (default 150, max 500), semantic (default false). The graph is built from up to 1000 of your most recent ACTIVE memories.
Performance
If the 3D map is slow, lower the Nodes budget. If WebGL is unavailable the page cannot render the 3D scene; use the 2D map in /app.