AI Code Security Scanner

Project Overview

A one-day prototype. A read-only scanner that reviews a codebase for security problems, built end to end in about a day.

Key Features

• Reviews a codebase for security problems and reports what it finds
• Read-only by construction: the agent's tools cannot write, so a scan can't modify what it's inspecting

Technical Implementation

FastAPI, Claude Agent SDK, Next.js.

Results

Tested against a purpose-built vulnerable app, it found all 6 planted vulnerabilities — plus 2 real ones nobody had planted — with zero false positives, in 70 to 100 seconds per scan.