Skill

domain-fintech:fraud-detection

Guides fraud detection implementation: rule-based detection (velocity checks, amount thresholds, geo-anomalies), ML-based anomaly detection (isolation forest, autoencoders), device fingerprinting, behavioral biometrics, 3DS2, fraud scoring pipelines, case management, and chargeback prevention. Use when building fraud prevention systems.

Install

npx claudepluginhub rnavarych/alpha-engineer --plugin domain-fintech

Tool Access

This skill is limited to using the following tools:

ReadGrepGlobBash

Preview

- Building or tuning velocity checks, amount anomaly rules, and geo-anomaly detection

Supporting Assets

references/3ds2-case-management.mdreferences/ml-scoring-pipeline.mdreferences/rule-based-detection.md

SKILL.md

Similar Skills

agent-harness-construction

Designs and optimizes AI agent action spaces, tool definitions, observation formats, error recovery, and context for higher task completion rates.

everything-claude-code

156.2k

accessibility

Designs, implements, and audits WCAG 2.2 AA accessible UIs for Web (ARIA/HTML5), iOS (SwiftUI traits), and Android (Compose semantics). Audits code for compliance gaps.

everything-claude-code

156.2k

agent-eval

Compares coding agents like Claude Code and Aider on custom YAML-defined codebase tasks using git worktrees, measuring pass rate, cost, time, and consistency.

everything-claude-code

156.2k

Stats

Parent Repo Stars9

Parent Repo Forks0

Last CommitMar 3, 2026

Actions

View Source View Plugin View on GitHub View README

domain-fintech:fraud-detection

When to use

Building or tuning velocity checks, amount anomaly rules, and geo-anomaly detection

Implementing ML-based anomaly detection (isolation forest, autoencoders)

Designing a real-time fraud scoring pipeline with combined rule + ML scores

Adding device fingerprinting or behavioral biometrics to authentication flows

Implementing 3DS2 and optimizing frictionless approval rates

Building fraud analyst case management queues and chargeback prevention

Core principles

Every false positive is a lost customer — tune thresholds per risk tier, not globally

Combine rules and ML — rules catch known patterns fast; ML catches novel behavior

Feature freshness determines accuracy — stale user profiles produce wrong scores at scale

Feedback loop is mandatory — case outcomes that don't retrain the model are wasted signal

Liability shift over blocking — 3DS2 frictionless approval with liability shift beats auto-decline

Reference Files

references/rule-based-detection.md — velocity checks, amount thresholds, geographic anomaly rules, rule engine schema with scoring, device fingerprinting signals and risk scoring

references/ml-scoring-pipeline.md — isolation forest and autoencoder models, model deployment (shadow/champion-challenger), behavioral biometrics, real-time scoring pipeline, scoring tiers, feature store design

references/3ds2-case-management.md — 3DS2 frictionless vs challenge flow, implementation via processor SDK, analyst review workflow, feedback loop, chargeback prevention and ratio monitoring

Fraud Detection

When to use

Building or tuning velocity checks, amount anomaly rules, and geo-anomaly detection
Implementing ML-based anomaly detection (isolation forest, autoencoders)
Designing a real-time fraud scoring pipeline with combined rule + ML scores
Adding device fingerprinting or behavioral biometrics to authentication flows
Implementing 3DS2 and optimizing frictionless approval rates
Building fraud analyst case management queues and chargeback prevention

Core principles

Every false positive is a lost customer — tune thresholds per risk tier, not globally
Combine rules and ML — rules catch known patterns fast; ML catches novel behavior
Feature freshness determines accuracy — stale user profiles produce wrong scores at scale
Feedback loop is mandatory — case outcomes that don't retrain the model are wasted signal
Liability shift over blocking — 3DS2 frictionless approval with liability shift beats auto-decline

Reference Files

references/rule-based-detection.md — velocity checks, amount thresholds, geographic anomaly rules, rule engine schema with scoring, device fingerprinting signals and risk scoring
references/ml-scoring-pipeline.md — isolation forest and autoencoder models, model deployment (shadow/champion-challenger), behavioral biometrics, real-time scoring pipeline, scoring tiers, feature store design
references/3ds2-case-management.md — 3DS2 frictionless vs challenge flow, implementation via processor SDK, analyst review workflow, feedback loop, chargeback prevention and ratio monitoring