USE CASE · DEFENSIVE REVIEW

Let AI review the entire project before a small issue becomes a large incident.

Project Brain expands review beyond one file: auth flows, configuration, permission checks, error handling, secret patterns, dependency usage and data paths.

Read-only audit

Start without write access to separate findings from remediation.

Evidence first

Every finding should include file/symbol/path evidence.

Fix + regression

Fix only approved findings, then run corresponding tests.

Use ChatCode with project web. Read-only. Defensively review auth/session/permission, input validation, error exposure, secret/config handling and remote/deploy scripts. Rank findings by severity, cite file/symbol evidence and verification steps. Do not edit yet.

No AI audit is a certification of “absolute security.” The output assists review; important systems still need testing, code review and an appropriate security process.