Skip to main content

Risk Scoring

Every domain and wallet processed by Chainara Platform receives a risk score from 0–99. This score is calculated by combining four independent weighted components, each measuring a different dimension of threat likelihood.

Score Ranges

ScoreLabelMeaning
0–24SafeNo significant threat signals detected
25–49Low RiskMinor concerns; worth monitoring
50–74Medium RiskSuspicious patterns detected; flag for review
75–89High RiskStrong indicators of malicious intent
90–99CriticalNear-certain threat; immediate action recommended

A score of 75+ will typically trigger an automatic flag in the Platform dashboard and appear in the threat feed.

The Four Scoring Components

The final score is computed as a weighted sum of four independent inputs:

ComponentWeightWhat It Measures
Rule Matches30%Configurable rule engine hits (keywords, patterns, TLDs)
Enrichment Signals25%DNS, RDAP, certificate, VirusTotal, URLhaus, etc.
LLM Analysis25%AI judgment on intent, content, and context
Threat Intelligence20%Known bad actors, blocklists, cross-platform correlations

Each component produces a sub-score of 0–100, which is then multiplied by its weight and summed:

final_score = (rules × 0.30) + (enrichment × 0.25) + (llm × 0.25) + (threat_intel × 0.20)

Rule Matches (30%)

The rule engine evaluates the domain or wallet against your configured ruleset. Each matching rule contributes a defined amount to this sub-score. Rules can be configured, enabled/disabled, and tuned by admins. See Domain Detection Rules for the full rule configuration reference.

Enrichment Signals (25%)

Nine enrichment modules run in parallel when a domain is processed, each contributing signals:

ModuleKey Signals
DNSMissing MX/SPF, suspicious nameservers, recent DNS changes
RDAP/WHOISDomain age (<30 days = high risk), privacy-masked registration
CertificateFree CA (Let's Encrypt), self-signed, certificate age mismatch
TyposquattingLevenshtein distance to known brands (ripple, xrp, coinbase)
VirusTotalReputation from 70+ security vendors
URLhausPresence in the URLhaus malware URL database
AbuseIPDBIP reputation and abuse history
GeolocationHigh-risk hosting countries, VPN/proxy detection
HTTP AnalysisPage content, redirect chains, suspicious JavaScript

Enrichment typically completes in 1.5–2 seconds (all modules run concurrently).

LLM Analysis (25%)

A large language model evaluates the domain holistically: examining the name, content, registration metadata, and enrichment signals together. The LLM provides:

  • A structured threat classification
  • Confidence level
  • A human-readable explanation of its reasoning

This component catches sophisticated attacks that evade pattern matching: for example, a legitimate-looking domain with subtle scam content buried in the page body.

Threat Intelligence (20%)

Cross-references the domain or wallet against:

  • Chainara's internal threat database (thousands of malicious domains and blacklisted wallets: see Platform Overview for current scale)
  • Community-submitted fraud reports
  • Active persona intelligence from the Intel platform
  • Historical IOC associations (linked wallets, related domains)

A direct match against a known threat immediately elevates this sub-score, even if other components show low scores.

Worked Example

A newly registered domain xrp-giveaway-official.live might score:

ComponentSub-ScoreWeighted
Rule Matches85 (keyword giveaway, TLD .live, xrp brand)25.5
Enrichment90 (domain age: 2 days, Let's Encrypt cert, privacy WHOIS)22.5
LLM Analysis95 (LLM identifies giveaway scam pattern)23.75
Threat Intel40 (no direct match yet)8.0
Final Score79.75 → 79 (truncated to integer)

Score 79 = High Risk: the domain would be auto-flagged and added to the threat feed. Final scores are always truncated (floor) to the nearest integer, never rounded.

Score Updates

Scores are not static. They are recalculated when:

  • New fraud reports link to the domain or wallet
  • Enrichment modules detect changes (e.g. domain content updated)
  • The Intel platform captures a new conversation referencing this address
  • An admin manually updates the classification

The Last Updated timestamp on each record shows when the score was last recalculated.

Using Scores in Workflows

ScoreRecommended Action
< 25No action required
25–49Add to watchlist; check again after 7 days
50–74Flag for manual review; alert security team
75–89Block or blacklist; submit fraud report
90–99Immediate blacklist; escalate to threat response

Scores can be used programmatically via the API: the risk_score field is returned on every wallet and domain lookup.