Auditor

Agentic security auditor — CLI + dashboard for code and LLM apps.

Deterministic security scanners (~40 rules across 7 categories) with optional LLM-enriched remediation, a CLI, and a Next.js dashboard.

Security auditing for code projects and LLM applications.

Component	What it does
Engine	Python package with 7 deterministic scanner categories (~40 rules): secrets, prompt injection, PII, authentication, insecure output handling, dangerous code, and configuration. OWASP/CWE knowledge base, optional LLM enrichment, and safe auto-fixes.
CLI	audit scan, audit explain, audit fix, audit info — CI-friendly exit codes and JSON output.
Dashboard	Next.js web UI over the same CLI — scan local projects, browse findings with remediation detail, shared history.
Demo	Intentionally vulnerable LLM chatbot with planted findings for end-to-end testing.

Fully functional offline; the LLM layer only adds deeper, code-specific remediation on top of the built-in knowledge base.

Stack: Python, Next.js, TypeScript, OWASP/CWE knowledge base.

Source: Utkarsh736/audit.