Skip to content

experiment_log.md

Latest commit

 

History

History
45 lines (33 loc) · 1.31 KB

experiment_log.md

File metadata and controls

45 lines (33 loc) · 1.31 KB
Raw

FileGuard Round 4 Experiment Log

Scope

Round 4 does not run long stability tests or require screenshots. It focuses on web-console completeness and system capability completion.

Demo Command

$env:PYTHONPATH='src'; python experiments/run_demo.py --config config.example.yaml

The demo writes only under experiments/sandbox/outputs/.

Covered Scenarios

  1. Normal file modification.
  2. Sensitive path and key-file policy hits.
  3. Alert escalation from repeated policy hits.
  4. High-entropy ransomware-like content replacement.
  5. Hash/fuzzy-hash change observation.
  6. Runtime incremental snapshot records.
  7. Dry-run defensive auto-restore candidate selection.
  8. Report trigger and API sample generation.

Artifacts

See experiments/sandbox/outputs/artifact_index.md after running the demo.

Verification

Required verification commands:

python -m compileall src tests experiments
$env:PYTHONPATH='src'; python -m unittest discover -s tests -v
$env:PYTHONPATH='src'; python experiments/run_demo.py --config config.example.yaml
$env:PYTHONPATH='src'; python -c "from fileguard.api.server import app; print(app.title, len(app.routes))"
cd frontend
npm install
npm run typecheck
npm run build
forbidden placeholder scan across src, tests, experiments, templates, frontend, and docs