HN Showcase
Back to browse
AI codebase analyser and auto-fixer

AI codebase analyser and auto-fixer

by nutifafa·Jul 25, 2026·1 point·0 comments
Visit ProjectView on HN

AI Analysis

●MidSolve My Problem

Regex-based static analyzer for Python AI apps when Bandit and Semgrep miss context.

Strengths
  • •Deterministic regex rules catch unsafe-tool-output patterns in milliseconds with zero LLM cost.
  • •Benchmark shows 100% recall on vulnerable fixtures where Bandit and Semgrep hit 20%.
Weaknesses
  • •Regex-only approach can't detect semantic vulnerabilities requiring code flow analysis.
  • •Optional LLM SAST pass via Ollama adds complexity for a tool marketed as 'no LLM call'.
Category
Security
Target Audience

Python developers building LLM-integrated applications

Similar To

Bandit · Semgrep · vulnhuntr

Post Description

The difference is the mindset shift of "human-in-the-loop" approach to debugging. I believe security is so critical that AI should never execute every action end to end.

Similar Projects

Developer Tools●●Solid

Detect Drift in TypeScript codebases

AST-based contracts beat raw code for AI, but codebase analyzers are increasingly crowded.

Niche GemBig Brain
AmiteK
104mo ago