Open
Conversation
Complete design specification for reconstructing MindGraph AI into a brain-like simulator implementing the Clustered Associative Recall (CAR) Protocol. Covers data model, 3D visual language (Three.js), 13-step retrieval animation, interaction design, and Claude ingestion manual.
Adds atmospheric layer (ambient particles, nebula clouds, star field), full interaction state coverage, emotional onboarding arc, anti-slop rules, design tokens (CSS variables), micro-interactions spec, accessibility (keyboard nav, screen reader, reduced motion), node placement algorithm, performance boundaries, and tooltip design.
Major spec updates: - Two-layer architecture: local Node.js server + browser frontend - Claude Code CLI as default AI runner (uses subscription auth) - Project mode: slides into existing projects via car-brain --project - Disk-based persistence (.car-brain/ directory) - Project ingestion pipeline with file scanning - Pluggable runner interface for future local model support - Updated .gitignore for .car-brain/ data directory
Replace node/edge types with CAR entities (chunk, cluster, question, pattern, trigger) and CAR relationships (linked_to, amends, contradicts, etc). Add 3D position support, design tokens, depth planes, retrieval step definitions, Ebbinghaus decay constants. Update graph-document with per-type metadata factories. Add cluster methods and retrieval state tracking to stores.
Replace all CSS with dark theme using spec design tokens (#080c14 canvas, #0d1117 panels). Anti-slop tabs (text-only, underline-active, IBM Plex Mono 10px uppercase). Add Three.js import map to index.html. Update title to CAR Brain. IBM Plex Sans + Mono fonts.
Replace DOM/SVG canvas with Three.js WebGL scene. Includes OrbitControls, UnrealBloomPass post-processing, FogExp2 depth fade. 3D node geometries per type (sphere T1, icosahedron T2, octahedron T3, tetrahedron questions, diamond triggers). Atmospheric layer with 200 ambient particles, 500 star points, nebula sprites, ambient light breathing. Camera intro swoop from (0,200,600) to home (0,80,350). Auto-focus on selected node with 0.6s easeInOutCubic lerp. Raycasting for node selection. Contradiction edge pulse animation. Legacy canvas sub-modules stubbed.
Rewrite top toolbar with memory input, query input, play/step/reset controls, and 13-step dot indicator. Rewrite left palette with 6 CAR tools (select, pan/orbit, add chunk, add link, add question, add trigger). Rewrite inspector with context-sensitive tabs per node type (chunk: overview/metadata/retrieval/timeline/scoring; question; cluster; pattern; trigger). Rewrite bottom panel with 5 CAR tabs (retrieval log, consolidation, contradictions, questions, metamemory) with badge counts.
Add relevance-scorer with Ebbinghaus decay curve and six weighted factors. Add question-generator with L1-L5 template-based generation. Add consolidation-engine for T1→T2→T3 tier promotion. Add contradiction-detector with keyword-based conflict detection. Add cluster-manager for auto-clustering by shared tags/entities. Add car-engine orchestrating the full 13-step retrieval sequence with play and step-through modes. Add retrieval-animator for visual coordination during retrieval.
Add Spain margins demo seed data (~15 chunks, 3 tiers, clusters, contradiction, trigger, questions). Create zero-dep Node.js local server (server.js) with REST API for disk persistence at ~/.car-brain/brains/. Update persistence-store for dual localStorage+server persistence. Rewrite app.js bootstrap with CAR engine wiring. Add CAR_INGESTION_PROTOCOL.md. Delete obsolete node components, inspector tabs, bottom panel views, old runtime modules, and sample data.
…atmosphere - Node radii 5x larger (1→5, 1.5→7, 2→9) so they're visible neurons not dots - Bloom strength 0.6→1.5, threshold 0.8→0.4 for dramatic glow halos - Emissive intensity 0.6→1.8 so nodes radiate light - Fog density reduced 0.0015→0.0008 for less depth fade - Ambient light 0.3→0.5 for brighter scene - Particle size 0.5→1.5, star size 0.3→1.0 with additive blending - Nebula sprites 200→400 scale for larger glow clouds - Edge opacity boosted 0.4→0.7 and dash sizes increased - Cluster shell radius 10→40 to properly enclose members
- Sprite-based node labels (canvas texture, white text with cyan glow) - Removed CSS2DRenderer (caused silent module crash on some browsers) - Increased ambient particles 200→400, stars 500→800, nebula 4→6 - Nebula gradients more vibrant (opacity 0.06→0.15 center) - Nebula sprites larger (500-700 scale) and wider spread - Reduced tone mapping exposure 1.2→0.85 for darker background - Added subtle grid floor for spatial reference - Node labels show first 18 chars of chunk name with ellipsis
Complete README covering architecture, 3D brain visualization, 13-step retrieval sequence, CAR protocol principles, project ingestion, persistence, tech stack, and keyboard shortcuts.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.