# EvidenceGene Court

- Page: https://scanovich.ai/projects/evidencegene-court/
- Language: en
- Updated: 2026-06-13

- Status: experiment
- Code: open source · MIT
- Repository: https://github.com/FUYOH666/evidencegene-court
- Research line: [Evidence and authority for machine decisions](https://scanovich.ai/research/evidence-and-authority.md)

Three AI agents — prosecutor, defender, arbiter — investigate disk and memory evidence. No finding enters the report without a reference to a real artifact.

Built for FIND EVIL! 2026, the SANS Institute hackathon for autonomous incident response.

## What works today

- Every tool is read-only and typed.
- Every claim is checked against a hash-chained audit log. A finding without a valid artifact reference is rejected by the interface, not by a prompt.
- The whole investigation runs on one laptop with a local model.

## Why it is in the lab

It is a small, public demonstration of the evidence-and-authority research: fabrication is refused by structure.

## Evidence

- **Accuracy report** — On a public forensic case the court identified the documented implant and rejected an injected fabricated finding. Measured 2026-06-13. Source: https://github.com/FUYOH666/evidencegene-court/blob/main/docs/ACCURACY_REPORT.md
- **Demo video** — The full investigation recorded on one laptop. Measured 2026-06-13. Source: https://youtu.be/HerCqv_LA3Q

## How to engage

The same structure for another kind of investigation where every claim must point at evidence.

[Contact](https://scanovich.ai/contact/) · private@scanovich.ai