Skip to content

README.md

Latest commit

 

History

History
79 lines (57 loc) · 2.62 KB

README.md

File metadata and controls

79 lines (57 loc) · 2.62 KB
Raw

Scripts

Helper and diagnostic scripts for the MalVol pipeline. Core stages (auto_vol.py, build_graph.py, etc.) live at the repository root.

IOC and corpus profiling

Script Purpose
ioc/code_injection_analysis.py malfind / injection IOC metrics
ioc/hidden_proc_analysis.py psscan vs pslist hidden process detection
ioc/filescan_analysis.py Suspicious file staging paths
ioc/network_analysis.py External connection / C2-style activity
ioc/analysis_corpus.py Aggregate IOC tables across corpus
ioc/upload_sessions_batch.py Batch session upload helper
corpus_profile.py Corpus-level graph/stat summary
vol3_model_scan.py Volatility 3 plugin scan utility

Dataset governance

Script Purpose
apply_hard_benign_labels.py Merge hard_benign_labels.csv into manifest (benign_subtype, train_eligible)

Training and calibration support

Script Purpose
calibrate_uncertainty_thresholds.py Fit gates → outputs/uncertainty_thresholds.json
fit_conformal_bundle.py Optional conformal review bundle for two-model path

Evaluation and diagnostics

Script Purpose
diagnose_triage_geometry.py Score overlap / margin report from analysis JSON
per_process_fp_report.py Per-process false-positive style breakdown
tabular_baselines.py Non-GNN baseline comparisons

Reporting

Script Purpose
generate_sample_reports.py Write REPORT.md per sample from analysis_report.json
export_dissertation_figure_data.py Export numeric series for external LaTeX figures (optional)

LaTeX helpers (optional)

Not required for core pipeline operation:

  • generate_analysis_latex.py
  • generate_eval_latex.py
  • generate_dissertation_figures.py
  • count_dissertation_words.sh

Tools

Script Purpose
tools/csvs.py CSV utilities

Typical workflows

# Per-sample markdown reports
python scripts/generate_sample_reports.py --base-dir extracted_csvs \
  --analysis-json outputs/two_model_analysis.json

# Threshold calibration after retrain
python scripts/calibrate_uncertainty_thresholds.py

# Geometry diagnostic
python scripts/diagnose_triage_geometry.py outputs/two_model_analysis.json

# Apply benign subtype overrides
python scripts/apply_hard_benign_labels.py extracted_csvs/dataset_manifest.csv \
  --labels-csv extracted_csvs/hard_benign_labels.csv

See docs/evaluation_operations.md and docs/Pipeline.md.