CodeSentinelAI explores an autonomous workflow that combines deterministic static analysis with LLM reasoning to detect, analyze, correct, and validate Python code.
Traditional linters can identify problems, while AI can reason about code. CodeSentinelAI explores bringing those capabilities together in a controlled loop.
Use Ruff and execution feedback to ground the auditing process in concrete diagnostics.
Use an LLM to interpret errors and propose context-aware corrections.
Don't stop at generating a patch. Check whether the proposed correction actually works.
The current development version uses a local LLM architecture and deterministic Python tooling.
CodeSentinelAI is an experimental project exploring AI-assisted software auditing and remediation.
Open the GitHub repository →