Skip to content

Rodex IDE

Rodex IDE - A Cursor-style IDE that uses 4 AI agents (powered by Nebuis Coding Model + Blaxel sandboxes) to perform real-time security analysis, bug detection, and autonomous fix application on Python codebases.

Code Video Cerebral Valley

Core trick

Four agents scan Python code, apply fixes in a sandbox, and accept a fix only if it compiles and removes the flagged pattern.

Tags

securitydev tools

Description

Rodex IDE - A Cursor-style IDE that uses 4 AI agents (powered by Nebuis Coding Model + Blaxel sandboxes) to perform real-time security analysis, bug detection, and autonomous fix application on Python codebases.
YuvrajGupta1808/Rodex-IDE★ 00 forksPythonMITCreated Apr 19, 2026 · last push Sep 6, 2026

README excerpt

# Multi-Agent Code Review IDE

[![CI](https://github.com/YuvrajGupta1808/Rodex-IDE/actions/workflows/ci.yml/badge.svg)](https://github.com/YuvrajGupta1808/Rodex-IDE/actions/workflows/ci.yml)
[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE)
[![Python 3.11+](https://img.shields.io/badge/python-3.11%2B-blue.svg)](https://www.python.org/downloads/)

A Cursor-style IDE where four AI agents perform security analysis, bug detection,
and autonomous fix application on Python code — streaming every step to the
browser over SSE. Fixes are not trusted on the model's word: each one must
compile *and* provably remove the flagged pattern before it is recorded.

Runs on **Gemini 2.5 Pro** (Vertex AI), **Gemini** (API key), or **OpenAI GPT-4o** —
the provider is one environment variable.

![Rodex reviewing a file: the Security agent flags SQL injection and unsafe deserialization, and the Fix agent patches them](docs/assets/rodex-demo.gif)

## Features

- **Cursor IDE UI** — Monaco editor, file tree, live agent panel, findings feed
- **4 Agents** — Coordinator, Security, Bug Detection, Fix/Patch
- **Real-time streaming** — SSE events stream all agent thoughts, tool calls, and findings live
- **Verified autonomous fixes** — every patch must compile *and* pass an AST check proving the flagged pattern is gone ([details](#fix-verification))
- **Per-agent telemetry** — tokens, reasoning tokens, latency, and USD cost per agent, in every review payload
- **Pluggable LLM p

Team