← back to Stable build & deploy plan
CTO Explainer · Source-Cited · No Assumptions

Trust Channels — From Compliance Primitive to General Infrastructure The exact recorded plan for generalizing trust channels beyond KYC/compliance into a universal trust-payment framework. Every claim cites a research source; status is marked PROVEN / PROPOSED / BLOCKED.

RESEARCH COMPOSITE 1.0 · 71 experiments DESIGN LOCKED · not implemented MAINNET-BLOCKED · legal + oracle
The one-line generalization: trust channels were re-cast from "is this party KYC-approved?" (binary, compliance) into trust-as-configuration — a programmable primitive where any two parties transact under scoped, earned, delegatable rules. Compliance becomes just one channel type among many, configured over the same stack as data marketplaces, agent payments, escrow, and subscriptions.
01

The core thesis

"Trust Channels abstract the trust relationship itself — allowing any two parties (human, agent, machine, enterprise) to transact under scoped rules without per-transaction negotiation."STRATEGIC_ANALYSIS_2026_03_29.md:9–16
"Trust is not binary. Every payment system today treats it as on/off (KYC or not, approved or not). Trust Channels make trust continuous, earned, delegatable, and composable. That's the zero-to-one."STRATEGIC_ANALYSIS_2026_03_29.md:14–15

The generalization axis, in the framework's own words: a shared smart-contract platform that supports multiple products through configuration — each product a "feature lane" sharing the same trust primitives, running multi-tenant, with an inheritable governance floor and a swappable settlement substrate. antilles-trust-framework-stable.html:118–120

02

Compliance becomes ONE channel type, not the center

The DomainRegistry / DomainFactory pattern makes "compliance" Domain 0 — the existing system — with new domains deployable for any use case. Reputation is earned through any channel activity, not just KYC attestations.

BEFORE — compliance-bound

Trust = a binary KYC/jurisdiction gate. One purpose: regulatory eligibility. Institutional only.

AFTER — general infrastructure

Trust = a continuous, earned score usable by any channel type. Compliance is one pluggable domain; data, agents, escrow, subscriptions are peers. DomainRegistry/Factory · 2026-02-17-sprint9-domain-trust-channels-design.md:8–24

On-chain, this rides existing mechanics: a rmtTrustChannelIndex selects which channel a route is checked against (getActiveTrustChannelRoutePossible(from, to, index)), and domain-scoped citation tracking gives "domain-filtered views of global PageRank." Compliance is simply index/domain 0. verified in ShyftGatedResolver / ShyftCacheGraph

03

The five general-purpose templates (none is compliance-only)

Recorded as the channel templates covering ~85% of expected Year-1 use cases. Each uses the same primitives — reputation gating, spending caps, session keys, pluggable "Rule 5" logic. STRATEGIC_ANALYSIS_2026_03_29.md:25–88 · q9-iter10:17–28

🔌 API Access
Sell API calls — data, AI inference, compute.
spendingCap · rateLimit · minRMT · per-call price
~2 min deploy · $500B API economy
🤖 Agent-to-Agent
An AI agent delegates payment authority to sub-agents.
sessionKey (DID) · spendingCap · allowlist · ttl · minRMT 200
~90 sec · $10B+ AI services
🗃️ Data Marketplace
Sell datasets, research, sensor streams.
accessTiers · expiryModel · minRMT 150 · attestOnPurchase
~3 min · $274B by 2028
🤝 Escrow / Milestone
Freelance, bounties, multi-party deliverables.
milestones · arbitrator · disputeWindow 72h · multi-sig release
~4 min · $455B freelance
🔁 Subscription
Recurring SaaS, content, retainers.
billingInterval · amount · grace · autoRenew · yield-escrow
~2 min · $1.5T SaaS
04

The generalized stack — compliance is an L4 plug-in

The framework separates configurable products (L4) from the shared trust engine (L2) and primitives (L1) over a swappable substrate (L0). antilles-trust-framework-stable.html:128–196

L4

Products & optional compliance plug-ins CONFIGURED, NOT BAKED IN

Compliance DeFi · identity portal · RMT dashboard · agent payment rails. Compliance = pluggable KYC/jurisdiction/allowlist add-ons over the shared stack.

L3

Agent-to-agent infrastructure GENERAL-PURPOSE

ERC-8004 agent registry + bridge · x402 trust channels (EIP-3009) · Stable Market. Not compliance-bound.

L2

Reputation & trust engine (RMT) — where compliance becomes one domain GENERALIZATION POINT

ReputationEngine + PageRankOracle + CitationCounters + DomainRegistry/Factory → per-domain trust channels. Scoped reputation: bots join multiple domains; compliance is Domain 0.

L1

Attestation & identity primitives SHYFT CORE

TrustAnchorManager/Storage · ShyftCacheGraph · ShyftConduit · ShyftGatedResolver (EAS↔Shyft) · tier model T0/T1/T2.

L0

Settlement substrate STABLE

USDT0-native EVM L1 (chainId 2201). Swappable (Base / OP-Stack on the locked multichain roadmap).

05

Proven vs proposed vs blocked (no overclaiming)

✅ PROVEN (1.0, 71 exp)
Sybil attacks economically irrational — all 3 attack scenarios −59% to −100% ROI. PROPOSAL_v2.md:12
Atomic agent participation — delegation + spending caps + reputation gating together; no incumbent does this. q9-iter10
Credit-score progression — 6-mo sim: payment history → RMT score → credit at ≥550.
5 templates + adoption mix validated in simulation.
📐 PROPOSED (design only)
On-chain credit lines (TrustCreditLine + Aave yield-backed). Architecture only. STRATEGIC_ANALYSIS §4
Multi-chain reputation (LayerZero/CCIP/Hyperlane). Not implemented. §3
Off-chain attestors (GitHub, LinkedIn, D&B, social) w/ ZK privacy. Design, no feeds. §9
Enterprise bulk onboarding template. Not validated with a real enterprise.
⛔ BLOCKED (pre-mainnet)
FM-1 Regulatory — 18 USC §1960 money-transmission classification; $200K legal opinion commissioned, not delivered. strategy.md:9 · PROPOSAL_v2.md:30–36
FM-2 Rule-5 malicious contracts — can drain escrow / bypass gates; mitigations designed, not deployed.
Oracle corruption ($300K) — needs 3-of-5 operators + delay buffer + challenger. Design pending. PROPOSAL_v2.md:65
06

Why it compounds — flywheel & network effects

more users → more volume → more attestations → higher RMT scores → better credit access → more users ↺
Datamore tx → better scoring → more creditworthy users
Liquiditymore escrow → better yield → more usage
Identity (moat)RMT history isn't portable; switching cost → ∞
Developermore channels → more templates → more devs
EnterpriseB2B integration pulls in client ecosystems
YEAR 1
~$180K
50K channels · $10M/mo vol
YEAR 2
~$1.8M
500K channels · $100M/mo
YEAR 3
~$72M
5M channels · $1B/mo + credit

Fee architecture: 0.1% of volume → treasury, split 0.05% oracle operators / 0.05% relayers; RMT 600+ users pay 0.02%. STRATEGIC_ANALYSIS:457–538projections are research models, not commitments.

Provenance. All claims traceable to recorded research: STRATEGIC_ANALYSIS_2026_03_29.md · PROPOSAL_FOR_X402_TRUST_CHANNELS_v2.md · strategy.md · FINAL_RESEARCH_SUMMARY.md · q9-iter10 results · 2026-02-17-sprint9-domain-trust-channels-design.md · antilles-trust-framework-stable.html. Cross-model review IDs (x402): grok 2c775b82 · gemini 1b63d56c · codex 8c329f7e.
Status honesty. Research composite 1.0 reflects research-eval, not production. The generalization is DESIGN-LOCKED and simulation-validated; it is NOT deployed and is mainnet-blocked on FM-1 (legal) + FM-2 + oracle hardening. Multi-chain, credit lines, and off-chain attestors are proposed architecture only.
Anything not in the sources was omitted, not invented.