Skip to content

Latest commit

 

History

History
255 lines (185 loc) · 11.1 KB

File metadata and controls

255 lines (185 loc) · 11.1 KB
OpenTrustEngine — open multi-dimensional trust scoring infrastructure

npm CI License Stars Bundle Website

Open trust scoring infrastructure for commercial reputation.

Replaces fragmented credit + platform reputations with a single, explainable 0–1200 score computed from real signals — payments, deliveries, disputes, reviews, KYC, accounting filings.
Built for lenders, marketplaces, government tenders, and B2B SaaS in India and emerging markets, where Plaid and FICO don't reach.

· Quick start · Architecture · Comparison · Use cases · Roadmap · Commercial ·


⚡ Quick start

npm install @ote/sdk
import { OpenTrustEngine } from '@ote/sdk';

const ote = new OpenTrustEngine({
  apiKey: process.env.OTE_API_KEY!,
  apiSecret: process.env.OTE_API_SECRET!,
});

// 1. Record a positive event
await ote.recordEvent({
  entityType: 'company',
  entityId: 'b1f2…',
  eventType: 'ESCROW_RELEASED',
  role: 'seller',
  rawValue: 5000,
});

// 2. Read the trust profile
const profile = await ote.getProfile('company', 'b1f2…');
//      ^? { company_score: 872, company_band: 'ABB', pillars: [...], ... }
sequenceDiagram
  autonumber
  participant App as Your app
  participant SDK as @ote/sdk
  participant API as api.opentrustengine.com
  App->>SDK: recordEvent({ entityId, eventType, role, rawValue })
  SDK->>API: POST /v1/events  (HMAC-signed)
  API-->>SDK: { delta: +6, newScore: 872, newBand: "ABB" }
  SDK-->>App: TrustEventResult
  Note over App,API: Score updates within 60 s · webhook fires on band change
Loading

Get a free API key at opentrustengine.com (no credit card).

See full SDK docs → · Embed the badge widget → · Connect Razorpay/Cashfree/Tally →


🏛 Architecture

OpenTrustEngine — system architecture diagram showing the flow from sources through connectors, the trust engine, the API surface, to consumers

The engine reads payments, deliveries, GST filings, reviews, and KYC events from any source via the connector framework, runs them through a 5-pillar Wilson + Beta + decay scoring model, and exposes the result through a public REST API + SDKs + an embeddable widget.

The math is intentionally conservative: small samples produce small scores (so cold-start entities cannot game the system), large samples drift toward their underlying quality.

Pillar Cap Signals
Transaction discipline 400 On-time dispatch, fulfilment SLA
Payment reliability 300 Escrow funding, on-time release, late %
Consistency & volume 250 Frequency, log-scaled value, partner breadth
Dispute resolution 150 Dispute rate, liable party, resolution speed
Peer feedback 100 Verified reviews & endorsements quality
Total 1200

Score ↔ band mapping (triple-letter, descending):

Band ≥ Score Tier Band ≥ Score Tier
AAA 1100 Elite BBC 500 Building
AAB 1000 Premier BCC 300 Growing
ABB 900 Trusted CCC 100 Starting
BBB 700 Reliable DDD 0 New

A composite Overall Trust Score (OTS) combines four sub-scores (work in progress, see Roadmap):

OTS = 1200 · ( 0.55·BTS̃ + 0.15·WTS̃ + 0.10·CTS̃ + 0.20·ITS̃ )

where BTS is business, WTS workforce, CTS compliance, ITS individual. The tilde ~ denotes range-normalisation.


🔍 Comparison

How OpenTrustEngine compares to existing reputation and credit scoring systems:

Capability OpenTrustEngine FICO SBSS eBay Feedback Lenddo Tala Schufa
Score range 0–1200 0–300 0–100 % 0–1000 300–850 100–600
Multi-dimensional (BTS+WTS+CTS+ITS) partial
Real-time updates (≤ 60 s) partial
Open public REST API + SDK
Open source ✓ AGPL
Self-hostable
Per-pillar explainability partial partial
Trust-bound digital signing
India / emerging markets focus n/a
Coverage of small businesses partial individual only individual only individual only

We're aware of newer alt-credit and identity providers (Sardine, Trulioo, Persona, Karza, Setu) and similar Indian fintech tooling. None of them are open source, none expose pillar-level explainability, and none bind a frozen trust snapshot into a legally-valid e-signature.


💼 Use cases

Lender (NBFC / co-op bank)

const profile = await ote.getProfile('company', borrowerId);
if (profile.company_score >= 700 && profile.company_band !== 'DDD') {
  approveLoan(borrowerId, computeRate(profile));   // ABB+ gets prime rate
}

Underwriters read BTS + CTS before approving a loan. Replaces ₹50,000 manual underwriting with a ₹500 API call. Loan-decision SLA from 7 days to 5 minutes.

Marketplace onboarding

const verify = await ote.verify('company', sellerId);
if (verify.verified && verify.band <= 'BBB') {
  fastTrackSeller(sellerId);     // skip manual KYC review
}

New sellers connect Razorpay + Tally during onboarding. OTE returns a band; marketplace skips manual KYC for sellers in BBB+. Drops onboarding from 7 days to 5 minutes for trusted sellers.

Government tender (GeM, IREPS) — TrustSign

A bidder submits a single TrustSign-signed credential as proof of past performance — the verifier reads the signer's frozen trust profile from the document via DRC. Eliminates the 30-page document submission per tender.

B2B SaaS (Tally, Zoho, etc.)

Embed @ote/widget in customer dashboards, deepen retention by exposing trust-score visibility. Co-marketing channel + sticky differentiator.


🛡 Why dual-license?

OpenTrustEngine is AGPL-3.0 for the open community and commercially licensed for closed-source / SaaS deployments without source disclosure.

  • Building on top of our hosted API at api.opentrustengine.com? No commercial licence needed — the SDK is a network client.
  • Forking the engine and self-hosting internally? AGPL is sufficient.
  • Embedding in a closed-source product or hosting modified code as a SaaS? Get a commercial licence →

This is the same model used by MongoDB, Elastic, Sentry, and Grafana.


🗺 Roadmap

Q2 2026 — open source launch (we are here)

  • Public REST API on api.opentrustengine.com
  • @ote/sdk, @ote/widget, @ote/connectors published
  • 6 connectors: Razorpay, Cashfree, PayU, Shopify, WooCommerce, Tally
  • TrustSign — Aadhaar eSign + DRC public verification (verify.onetrustengine.com)
  • AGPL + Commercial dual-license
  • Self-hostable Docker image with one-command bring-up
  • Self-serve dashboard at onetrustengine.com

Q3 2026 — multi-dimensional scoring

  • WTS — Workforce Trust Score (0–600)
  • CTS — Compliance Trust Score (0–400)
  • ITS — Individual Trust Score (0–600)
  • OTS composite formula in scoring orchestrator
  • React Native / Kotlin / Swift SDKs

Q4 2026 — verifiable credentials

  • W3C VC issuance for trust profiles
  • ZK-proof predicate verification (score ≥ τ without revealing exact score)
  • RBI account-aggregator framework integration
  • Research paper submission (IEEE TIFS / Computers & Security)

Beyond

  • Government tender pre-qualification module (GeM, IREPS)
  • Cross-platform identity stitching with provable Sybil resistance
  • Insurance pricing API (trade credit, performance bonds)

📦 Packages

Package Description
@ote/sdk TypeScript / JavaScript client for the OTE REST API
@ote/widget Embeddable trust badge widget — drop a <script> on any site
@ote/connectors Connector framework — Razorpay · Cashfree · PayU · Shopify · WooCommerce · Tally
@ote/shopify Shopify app for automatic seller scoring
@ote/tally-agent Electron desktop agent — sync Tally ERP into OTE
opentrustengine-woocommerce WordPress / WooCommerce plugin

🚀 Examples

Example What it shows
examples/basic-usage Record an event, read a profile, log result — 20 LOC
examples/next-app Next.js page with @ote/widget embedded
examples/webhook-server Express server receiving Razorpay webhooks via @ote/connectors

🤝 Contributing

PRs welcome. By contributing you agree to the CLA in CONTRIBUTING.md, which dual-licenses your contribution under AGPL-3.0 and the commercial license.

This is standard for projects with a commercial offering (MongoDB, Elastic, Sentry, Grafana, MariaDB, Tailscale).


📜 License

OpenTrustEngine is dual-licensed:

If you call our hosted API at api.opentrustengine.com from your application via the SDK, you do not need a commercial license — the SDK is a network client.

Copyright © 2026 Deepak Kumar Dwivedi, Freaquer.