thhepu/malicious-email-scorer

Add informative logger to assist with debugging

update tokenizer

Add dependencies and files for custom model support

Update requirements and files for custom models

Update README format for Hugging Face

Deploy modular architecture with custom LFS models

feat(api): implement modular layer toggles and empty-scan validation

fix(semantic): mitigate injection model saturation and redundant text parsing

feat(scoring): implement trusted sender reputation and cross-layer agreement gate

feat(l2): implement precision hardening and weighted scoring fusion

"chore: update gitignore to track evaluation datasets and pipeline scripts

small update

feat(api): implement production-ready FastAPI layer and Apps Script integration guide

feat(heuristics): implement Layer 1 foundation and orchestrator

test(heuristics): implement TDD test suite and evaluation scaffold for Layer 1

feat(semantic): implement Layer 2 semantic classifiers (Production Ready)

feat(layer1): implement Checks 4 & 5 with comprehensive security hardening This commit finalizes Layer 1 by introducing structural and metadata anomaly detection, followed by strict Red Team security hardening.

feat(heuristics): implement Layer 1 complete heuristic engines, orchestrator, and security hardening Implemented Check 1 (Hidden Content): 5 sub-detectors using Data-Driven Registry pattern for CSS, color camouflage, and ZWC.

feat(heuristics): implement Layer 1 foundation and orchestrator

test(heuristics): implement TDD test suite and evaluation scaffold for Layer 1