LeakLens
What is live
Filtering, expanding findings, switching fixtures, copying commands and replaying the recorded payload. Source-line highlighting follows the finding's real line number.
Read the evidence, not the badge
Seven real audits: two Python scripts, a corrected pipeline, two transaction tables, and two train/test splits. Every severity, evidence string and remediation below came from the LeakLens Python scanners at build time. The code examples also show the optional experimental AI hypotheses — confidence, local evidence and a remediation plan — in a separate panel that never changes the deterministic findings.
prerecorded No Python runs in your browser. GitHub Pages is static, so this page replays exact JSON generated by the Python scanner, dataset and remediation modules. For a live audit, run the command shown with each result, or connect a backend on the live scan page.
Loading recorded audit…
If this message stays, JavaScript is disabled; use the CLI commands in the quickstart.
Run the real scanner
python -m pip install leaklens-ml, then copy any command from a result. The demo's fixtures are regenerated with python scripts/build_site.py.
What it does not claim
The replay is not a new scan and the browser has no backend. Trust scores are computed centrally from findings by the CLI; the demo replays that value instead of recalculating it.