fix(F-14b): add pickle/model integrity checking mechanism

feat(retrain): add step8 — fit Platt after XGBoost training

feat(scripts): add fit_platt.py — standalone Platt calibration fitter

fix(train): record cv_method and feature_names in ensemble_results.json

fix(train): pass groups=cv_groups to cross_validate

fix: return None for corrupt images, add collate_skip_none, remove RandomGrayscale/RandomAffine (audit #10)

fix: use GroupKFold(source) instead of StratifiedKFold to prevent data leakage (audit #9)

fix: correct DATASETS dict, genimage->download_genimage_subset

fix: append missing DATASETS dict and __main__ block (truncation bug)

fix: append missing DATASETS dict and __main__ block (truncation bug)

fix: rewrite diffusiondb/genimage downloaders to use working sources

fix: restore missing DATASETS dict and __main__ block in download_ai.py

fix: remove trust_remote_code, use Parquet configs for DiffusionDB/GenImage

fix: use shap.Explainer for XGBoost 2.x compatibility

fix: exclude 'source' column from XGBoost feature matrix

fix: rewrite build_clip_database.py to read from manifest.csv

fix: lower image size filter from 64px to 16px minimum

fix: use as_posix() for manifest paths to avoid Windows backslash bug

fix: replace unicode box-drawing chars that crash Windows cp1252

feat: add extract_features + retrain_all pipeline scripts

fix(test+frontend+scripts): methods count, 25->26, logo, script cleanup

feat(ops): add production readiness checklist script

fix(accuracy): extract features from train split not val split

feat(ml): evaluation pipeline and improved XGBoost training from diagnostic report

Merge branch 'main' into feature/platform-forensics

feat(platform): invariant tests + pyflakes cleanup

feat(ensemble): — XGBoost trained classifier replaces hardcoded signal weights

fix: audit pass — unused imports, None crash guard, emojis, accuracy claims, broken f-strings

feat: complete — trained EfficientNet detector, centroids, integrated into ensemble

feat: add EfficientNet training script — reads manifest.csv, trains binary classifier

feat: add dataset download scripts for real and AI sources

Build CLIP reference database (#32)

Add reference dataset download scripts (#32)