The inner & embodied self —
observed, fractionalised, patterned by choice, externalised, then re-embodied.

Aura Engine is a player-built course where you generate your own data, then assemble Mixture-of-Experts (MoE) pipelines to choreograph your days. Everything is an atom. Explode, slice, fuse, and hot-swap your life in microticks — with federated privacy and consent on chain.

Get Started Roadmap

What is it?

A local-first, open-source website + starter stack where learners drag pre-categorised atoms (5–15 minute units) into days/weeks; the scheduler protects fixed blocks (e.g., HBOT 120m) and fills remaining microticks with context-aware work, wellness, XR, and reflection.

Pre-categorised library (Core + Electives)
DnD timeline composer
MoE agent DAG
Dimensional data stores (tabular + vector + blobs)
SIWE + DID, encrypted exports

Mantra

Coach people to fractionalise their own data and synchronise it across modalities for GenAI — with Web3 trust and federated privacy. That turns life into a game that actually respects the player.

Design principles

Everything is an atom Explode / Slice / Fuse Interrupt-friendly scheduling Consent-first, defaults private Readable JSON everywhere

Architecture overview

Frontend

Next.js / React, Tailwind, shadcn/ui, Framer Motion. Timeline composer with DnD; DAW-like rows per modality; copyable JSON.

Data Plane

Postgres (tabular), Qdrant (vector), IPFS/S3 (blobs). Local-first CRDT cache; one-click encrypted export.

Agents & MoE

normalise → schedule → culture → place → bio → xr → recommender → reflect. All agents share a simple JSON contract.

Agent I/O (TypeScript interface)

// Agent contract
type FeaturePack = Record<string, unknown>;
type AgentInput = { features: FeaturePack, context?: Record<string,unknown> };
type AgentOutput = { events?: any[]; recommendations?: any[]; artifacts?: any[] };
type AgentFn = (input: AgentInput) => Promise<AgentOutput>;

Atoms & Modules

Atoms are 5–15 minute micro-units (communication cues, place checks, breath sets, journal prompts, XR steps). Modules are curated bundles of atoms.

Atom (JSON)

{
  "id": "culture.greetings",
  "title": "Greetings & Introductions",
  "intention": ["community", "appropriateness"],
  "type": "atom",
  "time_block": "micro",
  "format": ["mental", "field"],
  "location": ["urban", "nature", "home", "digital"],
  "outputs": ["etiquette_checklist.md"],
  "filters": ["logical", "emotional", "rational"],
  "hooks": ["feature_store:culture", "agent:culture", "scheduler"],
  "privacy": "private"
}

Module (bundle)

{
  "id": "module.lesser_understood_place",
  "title": "Visit a Lesser-Understood Place",
  "type": "core",
  "atoms": [
    "culture.greetings",
    "culture.body_language",
    "place.local_table_pull",
    "time.sun_moon_align",
    "social.action_planner"
  ],
  "suggested_duration": "short",
  "outputs": ["location_brief.pdf", "consent_notes.json"]
}

Pipelines (player-built)

Learners wire their own DAG: Intake → Normalise → Feature Store → Agents → Composer → Calendar → Reflection.

Pipeline Preset (JSON)

{
  "name": "my_directional_pipeline_v1",
  "nodes": [
    {"id":"intake", "type":"collector", "sources":["journal","voice","xr","bio"]},
    {"id":"normaliser", "type":"agent.normalise", "in":["intake"], "out":["features"]},
    {"id":"features", "type":"store.vector+tabular"},
    {"id":"scheduler", "type":"agent.schedule", "in":["features"]},
    {"id":"culture",   "type":"agent.culture",   "in":["features"]},
    {"id":"place",     "type":"agent.place",     "in":["features"]},
    {"id":"bio",       "type":"agent.bio",       "in":["features"]},
    {"id":"composer",  "type":"agent.compose",   "in":["scheduler","culture","place","bio"]},
    {"id":"reflect",   "type":"agent.reflect",   "in":["features","composer"]}
  ],
  "ticks": { "size_minutes": 10, "guardrails": ["HBOT_120m_daily"] },
  "signals": ["satisfaction","fatigue","context_switch_cost"],
  "consent": { "share": "none" }
}

Microtick Scheduler (demo)

A tiny simulation to visualise available minutes after reserving your daily HBOT block.

Game frequency Microtick (min)

HBOT reserved: 120 min/day Working window: 2 h/day Available microticks: 0

Available

Used/Reserved

How it works

Select a frequency and microtick size. The demo assumes a daily planning window of modeHours/day, then subtracts the fixed HBOT block (120 min). It shows remaining microticks for atoms (5–15 min units). The real engine also applies location/device constraints, priorities, and context-switch costs.

Casual 2h/day • Busy 4h/day • Hyper 6h/day
Part-Time 3d/wk@4h • Full-Time 6d/wk@4h • Immersion 7d/wk@6h

Privacy, Trust, Federation

Local-first by default

CRDT docs remain on device; opt-in sync to your personal node. One-click encrypted export (JSON/CSV/media/IPFS CIDs). No dark patterns.

Web3 consent

Sign-In with Ethereum (SIWE) + DID. On-chain ConsentRegistry logs hashed grants/revokes. Non-transferable SBTs prove authored Module Decks and Pipeline Presets without leaking data.

Get started

Repo bootstrap

git clone <your-repo> aura-engine
cd aura-engine
# Web app (Next.js) would live in /apps/web
# This static file is the stepping stone landing.

Batch-1 (first week)

Choose a frequency preset; reserve HBOT 120m daily.
Each day: 1 core module + 1–2 micro atoms.
End-of-day: reflection (text/voice) → local features.
Weekend: assemble your first pipeline preset; run a 3-day sim.

Roadmap

Phase 1 — MVP

Static library, simple scheduler, encrypted exports, this site.

Phase 2 — Agent DAG

Normaliser, vector store, recommendations, calendar composer.

Phase 3 — Federation

Personal nodes, consent registry, SBT proofs, community packs.