TerAustralis Incognita
The Archive · TerAustralis Incognita

Crystal Core — Full Stack Blueprint v0.2 → v0.3

Decode · Ingest · Upgrade

Layer label: VISION with engineering targets. This is the aspirational full-stack blueprint for the Crystal universe, captured as canon. For what is actually built and running today, see ARCHITECTURE.md — the two documents must never be confused.

Single sentence: A sovereign mesh where nodes own data, a digital twin meters real flows, credits pay for services, CORE aligns incentives, GitHub + chain govern change — with Decode / Ingest / Upgrade as the data lifecycle.


0. DECODE (thread → system role)

Thread Decoded meaning System role
CrystalCore Sovereign kernel / reactor Protocol + runtime identity, stake, receipts
Crystal Vision Shared sight Twin UI, layered flows, public specs
TerAustralis Incognita Uncharted frontier Expansion regions, pilot deployments, epoch narrative
Seven Sisters spine Mythic geometry + story epochs Federation corridors, release calendar (cultural governance required)
4 layers Mythic → Physical → Economic → Governance Bounded contexts in architecture
Burn-and-mint + credits Stable UX + token incentives Payments + settlement service
RFC-001 Identity + ServiceReceipt Crypto + metering contract
Starline Budapest Reference physical node Mobility/energy twin + launch template for hubs

1. INGEST (unified system map)

Experience (Vision UI · Starline hub · mobile) → Edge (libp2p node · SQLite/CRDT · sensors · agents) → DECODE → INGEST → TWIN → RECEIPT engine → ECON burn/mint → Governance (GitHub RFCs · sim reports · on-chain params · slash), with an UPGRADE service closing the loop.

2. UPGRADE (v0.1 → v0.2)

Area v0.1 v0.2
Data path Implied metering Explicit Decode → Ingest → Twin → Receipt
Identity RFC draft did:crystal + agent sub-DIDs + capability macaroons
Economics Formulas parameters.yaml + batch receipt roots on-chain
Mobility Starline visuals Starline Budapest as canonical pilot region pack
Mythic Metaphor Epoch gates: no production mythic strings without council flag
Ops Phases Monorepo + K8s + one Helm chart per region
AI nodes Open question Agent bond + scoped receipts, parent liability DID

3. Full stack (by layer)

  • Clients: Twin web UI (layered flows, deck.gl/MapLibre), mobile offline-first, Tauri operator desktop, shared design tokens (navy / cyan / gold).
  • Edge & mesh: libp2p (QUIC, gossipsub), SQLite + Litestream, Automerge CRDT, Rust/Go node agent, OPA policy sidecar.
  • Core services: api-gateway · decode (schema/units/dedupe/fraud) · ingest (stream, partitioning, backpressure) · twin-api (PostGIS + TimescaleDB) · receipt-engine (RFC-001, dual-sig, Merkle batch roots) · econ (double-entry credits + chain worker) · governance-api · upgrade (governance-gated GitOps) · search.
  • Digital twin: H3 spatial index, TimescaleDB series (water, energy, egress, GPU-s), CloudEvents → twin.event/1, IPFS checkpoints, Flink simulation (later).
  • Economic: app-chain (Cosmos SDK or OP Stack L2), stake/slash/ReceiptRoot/ BurnMint/treasury contracts, off-chain credits ledger, cadCAD/Mesa sims.
  • Governance: GitHub RFCs + CODEOWNERS; on-chain votes/timelock; auditor guild attestations (IPFS CID); off-chain cultural council charter.
  • Observability & security: Prometheus/Grafana, Loki, OTel, Vault/KMS, GitHub Actions → Argo CD, K8s cloud + K3s edge.

4. Pipeline service specs

DECODE

Input: raw mesh frames, MQTT, HTTP, file drops → Output: crystal.twin.event/1; invalid → quarantine topic.

{
  "schema": "crystal.twin.event/1",
  "event_id": "uuid-v7",
  "source_did": "did:crystal:...",
  "h3": "8abe...",
  "class": "energy.kwh",
  "value": "12.4",
  "unit": "kWh",
  "observed_at": "ISO8601",
  "raw_ref": "bafy..."
}

Rules: JSON Schema, unit conversion, monotonic meter checks, replay window.

INGEST

Partition by h3 + class, write time-series store, emit receipt windows, fan out to Vision WebSocket. SLA: p95 < 2s cloud path; 72h edge offline buffer.

UPGRADE

Input: approved RFC + on-chain timelock → rolling node updates, schema migrations, epoch parameter push, blue/green API. Hard rule: economic param changes require an attached sim report artifact CID.

5–8. Monorepo, API, Budapest, build order

Monorepo: apps/ (vision-web, mobile, operator, starline-budapest) · services/ · node/crystal-agent · chain/ · packages/ (crypto, receipts, schemas, ui-tokens) · sim/ · spec/rfc/ · deploy/helm + regions + argocd.

MVP API: POST /v1/decode/preview · POST /v1/ingest/events · GET /v1/twin/flows?h3=&layer= · POST /v1/receipts + /confirm · POST /v1/econ/burn · GET /v1/econ/credits · GET /v1/gov/rfc · POST /v1/upgrade/epochs/{id}/apply. Auth: node DID challenge + capability macaroon; humans via OIDC → wallet link.

Starline Budapest pilot: 3–10 edge operator nodes; region cloud runs decode/ingest/ twin/vision; first receipt class mobility.checkin or energy.kwh at hub.

Build order: S1 crypto+receipts → S2 agent gossip → S3 decode+ingest+store → S4 twin+UI → S5 receipts → S6 testnet econ+sim → S7 upgrade+GitOps+Budapest Helm → S8 security audit gate → pilot.


v0.3 — Gap fill (the skin, nerves, and legal skeleton)

Honest checklist of what v0.2 lacked:

Gap Why it matters
Threat model & trust zones Sovereign mesh = attack surface at every node
Human journeys No onboarding = no network
Fiat / payments / treasury Credits need real rails
Compliance Energy, water, money, PII = licenses
Interop standards Utilities don't speak "custom JSON"
Federation & charters "No hierarchy" still needs rules to join
Data lifecycle & erasure Sovereignty = delete/export, not slogans
AI agents (full subsystem) Bond, liability, receipts, rate limits
Dispute & slash courts Economics without arbitration = griefing
SRE: SLO, DR, chaos Pilots fail on ops, not whiteboards
DX: local dev, SDK, sandboxes Builders need a 15-minute start
Testing pyramid Contract, sim, load, Byzantine drills
Schema registry & versioning Upgrade path must be explicit
Privacy / selective disclosure Twin data is sensitive
Cultural governance (operational) Seven Sisters can't be decoration
Product ↔ brand map Vision vs Core vs TerAustralis vs Starline
Commercial model Who pays whom, when, in what jurisdiction

A. Product map: TerAustralis = holding narrative · Crystal Vision = product UX · Crystal Core = protocol/runtime/economics · Starline Budapest = reference region pack · 21st Europe Starline = external reference, integrate via partner/API, never rebrand.

B. Trust zones: Z0 device (user keys) → Z1 edge mesh (peer-attested) → Z2 region cloud (charter) → Z3 global (governance) → Z4 partners (contractual). Threats & answers: sybil → stake + peer diversity; fake receipts → dual-sig + auditor lottery + meter hash chain; collusion mint → caps + anomaly detection; slash griefing → dispute bonds + appeals; key theft → hardware + rotation + short-lived macaroons; eclipse → multi-bootstrap + IPFS checkpoints; governance capture → stake caps + timelock; AI-agent abuse → sub-DID bond + scoped caps + parent slash liability; cultural harm → council veto on mythic assets.

C. Human journeys: citizen (install → credits → grant capability → consume → export/erase) · operator (keys → stake → run agent → earn mint → unbond 14d) · builder (clone → make dev → RFC PR with sim → staged deploy) · auditor (beacon selection → evidence CIDs → attestation → mint release or slash ticket).

D. Federation charters: signed charter doc (IPFS CID) + stake threshold; per-class min stake, auditor set size, allowed receipt classes, mythic epoch only if council_approved: true, named dispute court. Budapest = first charter template.

E. Integrations: energy IEC 61850/Modbus/MQTT · water SCADA/OPC-UA · mobility OCPI/GTFS-rt · payments PSP webhooks · gov OpenAPI adapters · GitHub webhooks · identity OIDC/eIDAS. Adapter contract: all adapters emit crystal.twin.event/1 only — never bypass decode.

F. Payments & treasury: fiat→credits via PSP webhook; CORE burn→credits via indexer; operator payout via mint + optional off-ramp; enterprise invoices via treasury multisig; nightly PSP⟷ledger⟷chain reconciliation. Legal note: token + credits may trigger MiCA / e-money rules — engage counsel before any public sale.

G. Data lifecycle / GDPR: export = signed archive of twin+receipt CIDs; erase = revoke macaroons + tombstone PII (chain stays hash-only); retention per class (raw 90d, aggregates 7y, charter-tunable); minimization = DIDs + hashes, never names; sovereignty default = primary copy on node, cloud replica opt-in.

H. AI agents: identity did:crystal:parent#agent-N; parent-staked slashable bond; macaroon-scoped classes and daily caps, no governance keys; parent co-sign for mint; parent liability on fraud; per-agent rate limits in decode.

I. Dispute court: bonded dispute within window → evidence packet (receipt + twin CIDs + auditor report) → markdown docket court repo → release/slash/ban → one appeal at higher quorum. GitHub is process; chain executes after timelock.

J. SRE: ingest 99.5% MVP / 99.9% prod; decode→twin p95 < 2s; RPO 1h / RTO 4h; Litestream + PG PITR; secondary-region DR (mesh survives cloud loss); chaos drills (kill ingest, 72h mesh buffer); public status page; per-region FinOps.

K. DX (15-minute start): git clone && make dev && make seed-budapest && make test && make sim; @crystal/sdk (TS), @crystal/cli, sandbox testnet with faucet, Pact contract tests between decode ↔ ingest ↔ twin.

L. Testing pyramid: unit (crypto/receipts/schema) → contract (OpenAPI+Pact) → integration (compose e2e) → sim (cadCAD sybil/bear) → load (k6) → Byzantine (malicious libp2p nodes) → red team. Gate: no mainnet without external audit + chaos weekend.

M. Schema registry: packages/schemas/ with registry.json compatibility matrix; breaking change → new major + migration job; nodes advertise supported_schemas[]; decode rejects unknown majors unless epoch-flagged.

N. Privacy roadmap: MVP hashes+DIDs → v1 SD-JWT selective disclosure → v2 ZK proof-of-SLA research spike.

O. Cultural governance (operational): MYTHIC_SPINE.md (core draft) + CULTURAL_PROTOCOL.md (council approval required); CI check on council_approved: true before any epoch name ships; block list — no sacred secret material, no unapproved Songline commercial use. Seven Sisters in software: only as epoch codenames inside approved charters.

P. Commercial & org: revenue = credit-sale margin (treasury+ops), burn fees (protocol/stakers), enterprise contracts (60% operators / 25% buy-burn / 15% treasury), grants via RFC awards. Phase gates: Seed = mesh MVP + RFC-001 + sim published; Series A = Budapest pilot 50 nodes + audited testnet; Scale = second federation + licensed payments.

R. Monorepo additions: integrations/ · services/{dispute-court,agent-runtime, payments-ledger,schema-registry} · security/THREAT_MODEL.md · compliance/{GDPR_ROPA,LICENSING_CHECKLIST}.md · courts/templates/ · sdk/{typescript,cli} · tests/{e2e,byzantine,load}.

S. Toward v0.4: full OpenAPI 3.1 · compose+Makefile scaffold · RFC-002 governance/ upgrade binding · RFC-003 federation charter + dispute · Budapest sample data (1000 events + 50 receipts) · MiCA legal memo outline · sequence diagrams (burn, mint, slash, launch day).


Blueprint captured as canon · VISION labeled · the runnable spine lives in services/

TerAustralis Incognita acknowledges the Traditional Custodians of the lands, waters and skies across Australia, and pays respect to Elders past and present. Sovereignty was never ceded.

Code and content licensed under CC BY-NC-ND 4.0. ABN 70 741 068 059. © 2026 TerAustralis Incognita. TerAustralis Incognita™ and CrystalCore™ are unregistered trade marks; no licence here grants any trade mark right.

Listen to the mythos soundtrack on Suno.

Interactive demos (simulated data, Authority HOLD): the operator shell and the citizen shell.

Evidence-first route health: the shared footer audit.

Non Solus — Not Alone