How imayankss/secureswipe works
SecureSwipe is a defense-only fraud-risk manager: it ranks suspicious transactions against a human-review capacity and returns bounded review outcomes, never payment authorization or decline decisions. Its reference API supports verified-bundle scoring, provenance, audit and idempotent replay; the browser also presents product, evidence and demo surfaces. Offline data preparation, model evaluation and monitoring preserve the scientific evidence behind the capacity trade-off. The README documents the workflow and evidence boundaries; several subsystem paths below come from the tree and index, so their internal calls are not inferred.
Main entry points: main.py Reference API · service.py Model service · bundle.py Verified bundles · batch_scoring.py Batch scoring
Components 17
Reviewer surfaces
- Product surface
web/app/page.tsx - Evidence surface
web/app/evidence/page.tsx - Bounded demo
web/components/product/BoundedReviewWorkflow.tsx
Bounded serving
- Reference API
api/main.py - Request schemas
api/schemas.py - Model service
api/service.py - Audit receipts
api/audit.py - Replay store
api/postgres_idempotency.py
Offline evidence
- Historical quarantine
src/data/historical_quarantine.py - Reference data
src/data/historical_reference.py - Capacity evaluation
src/lane_a - Offline monitoring
src/monitoring/offline.py - Model explainability
src/explainability/shap_explainer.py
Model internals
- Verified bundles
src/artifacts/bundle.py - Batch scoring
src/inference/batch_scoring.py - Feature preprocessing
src/preprocessing/preprocessors.py
Other components
- Risk analyst
Connections 17
- Risk analyst → Product surface: reviews
README.md - Risk analyst → Evidence surface: inspects evidence
README.md - Risk analyst → Bounded demo: runs walkthrough
README.md - Reference API → Model service: dispatches inference
api/main.py - Reference API → Request schemas: validates requests
api/main.py - Reference API → Audit receipts: records decisions
api/main.py - Reference API → Replay store: replays requests
api/main.py - Model service → Verified bundles: verifies bundle
api/service.py - Model service → Batch scoring: scores transactions
api/service.py - Batch scoring → Feature preprocessing: transforms features
src/inference/batch_scoring.py - Model service → Request schemas: returns predictions
api/service.py - Verified bundles → Feature preprocessing: records feature contract
src/artifacts/bundle.py - Risk analyst → Capacity evaluation: reviews capacity results
README.md - Risk analyst → Model explainability: inspects explanations
- Risk analyst → Offline monitoring: reviews monitoring
- Risk analyst → Reference data: reviews reference evidence
- Risk analyst → Historical quarantine: reviews quarantine evidence