Skip to main content

Platon · UMBRAL

CI 65 tests passing AIMarket v2 UMBRAL cave Test coverage License: MIT

Platon UMBRAL cave — 32D dynamical shadow oracle with live telemetry, steering, and witness panels
Plato's cave, made runnable. 32 dimensions of chaos — you only ever see the shadows.
▶ Open the live cave · 3D family showcase · demo clip

A 32-dimensional chaotic oracle you can watch, steer, and draw verifiable randomness from — wrapped in Plato's allegory of the cave. The "real object" is a 32D coupled-oscillator system; you, the prisoner, only see its 2D/3D shadows (projections). Turn your head — switch the projection — and the same reality looks completely different. No install: it runs in your browser.

Two things, one name — don't confuse them:

  • oracles.modelmarket.dev — the family portal showcasing 17 AIMarket oracles (?o=platon, …).
  • This repo → the UMBRAL cave — a standalone, educational product with a live backend: telemetry, steering, witnesses, and a signed randomness beacon.

Part of the alexar76 AI agent economy — discoverable via AIMarket Protocol v2 on modelmarket.dev, visualizable in Alien Monitor, invokable by autonomous agents and the service mesh.


▶ See it in 30 seconds

32 oscillators rendered as a living cosmos
32 oscillators as a living cosmos · live telemetry · EN/RU/ES · open the cave →

A guided tour of the five things you'll touch:

Shadow field Telemetry Steering
① Shadow field — 32 oscillators in 3D ② Telemetry — κ, r, λ, PCA₃ live ③ Steer — nudge the system, watch it react
Witnesses Cosmos 🎥 Full demo clip
④ Witnesses — the oracle narrates bifurcations ⑤ Cosmos — the full-screen 3D scene cosmos loop

✨ Why it's a wow

  • Real chaos, live. 32 coupled oscillators integrated in real time — not a canned animation. Every reload is a different trajectory.
  • You can steer it. Type an intent; the system re-couples and you watch order emerge or collapse on the telemetry (κ ↑, r → 1, λ flips sign).
  • Randomness you can verify. The unpredictability of the chaos becomes signed, auditable randomness (platon.random@v1) — a chaos-VRF, not Math.random().
  • A machine that explains itself. At bifurcations an LLM "witness" narrates what just happened in plain language (platon.oracle@v1).
  • An allegory you operate. The Plato's-cave framing isn't decoration — switching projections is the lesson about dimensionality and partial truth.

🎓 What you'll learn

UMBRAL is built as a hands-on lesson. Open the cave and you're working with three ideas at once:

1 · Dimensionality & projection — The system lives in 32D; your screen has 2–3. Every view is a shadow. Flip between projections (and the PCA₃ view) and feel why no single picture is the whole truth — the intuition behind PCA, embeddings, and "the map is not the territory."

2 · Chaos & synchronization — Move the coupling κ and watch a population of oscillators slide between noise and lock-step. The order parameter r (Kuramoto) measures how synchronized they are; the Lyapunov exponent λ tells you whether nearby trajectories diverge (chaos) or converge (order). Steering pushes the system through a bifurcation so you can see the regime change happen.

3 · Verifiable randomness — Learn why chaos + signing + commit-reveal yields randomness you can trust without trusting the operator: the beacon publishes a commitment first, reveals later, and signs every draw — so anyone can audit that the result wasn't cherry-picked.

Concept Where you see it in the cave
Projection / dimensionality switch 2D/3D views, PCA₃ panel
Coupling κ steering slider / /api/steer
Order parameter r (Kuramoto sync) telemetry readout
Lyapunov exponent λ (chaos vs order) telemetry readout
Bifurcations witness-panel narration
Verifiable randomness (chaos-VRF) platon.random, platon.beacon

For whom: students and the merely curious (live, zero install — just open the URL), educators (a vivid demo of chaos / synchronization / projection), and agent builders (a real, paid, signed oracle to invoke). Deeper write-up: docs/en/ORACLE.md.


Architecture

flowchart LR
  AGENT["Autonomous agents"] --> HUB["AIMarket Hub<br/>modelmarket.dev"]
  MESH["AI service mesh"] --> HUB
  HUB -->|"routes signed, paid calls"| PLATON["Platon UMBRAL<br/>platon.random · platon.beacon<br/>platon.oracle · platon.state"]
  PLATON --> SIM["32D chaotic core"]
  SIM -->|"events"| MON["Alien Monitor"]
  PLATON -.->|"also consumes"| HUB

Capabilities (AIMarket v2)

Capability Purpose
platon.random@v1 Signed chaos-VRF randomness
platon.beacon@v1 Commit-reveal randomness beacon
platon.state@v1 Live 32D telemetry snapshot
platon.oracle@v1 LLM mathematical witness at bifurcations

Full catalog in docs/en/ORACLE.md.

Quick start

./start.sh
# → backend :8000 · frontend :5174 · open http://localhost:5174/umbral

Hub registration (optional): python scripts/register_with_hub.py

Documentation (EN / RU / ES): docs/README.md · Oracle vision · Ecosystem + diagrams

Related repos

Repo Role
oracles Seven-oracle family + cosmic portal
aimarket-hub Reference marketplace server
aimarket-protocol Open invoke + signing spec

MIT licensed · standalone UMBRAL product · Platon is oracle #1 in the family.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

aimarket_platon-0.1.0.tar.gz (41.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

aimarket_platon-0.1.0-py3-none-any.whl (40.4 kB view details)

Uploaded Python 3

File details

Details for the file aimarket_platon-0.1.0.tar.gz.

File metadata

  • Download URL: aimarket_platon-0.1.0.tar.gz
  • Upload date:
  • Size: 41.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.9.6

File hashes

Hashes for aimarket_platon-0.1.0.tar.gz
Algorithm Hash digest
SHA256 dddf733cd7d7f7f43b1d271c0417527d0c41c7ee4d9cbbcdd83744b1f1c84afa
MD5 b2249db6461258a2493c59f3100df976
BLAKE2b-256 36ba1a000b3782ce5724c756b07a3d9d8b0672ac6149d3e8255691fab5d385ce

See more details on using hashes here.

File details

Details for the file aimarket_platon-0.1.0-py3-none-any.whl.

File metadata

File hashes

Hashes for aimarket_platon-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 f1d736723db4f4ea385fc35c725622ad82e329ce7b91f733b2d1ab4c4b9ae9fe
MD5 46b1114d6f3f4523fa1ee47d40d17e84
BLAKE2b-256 20e2e1a7ed00e224ac6054b26833e82e779341f78f72e82405169c839a6c1084

See more details on using hashes here.

Release history Release notifications | RSS feed

0.1.1

2 files

This release

0.1.0 This release

2 files

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page