Neo on Cursor

Quick TL;DR

Neo brings specialized AI and ML engineering capabilities to Cursor, including data pipelines, experiment tracking, model deployment, and autonomous debugging, all on top of Cursor’s built-in AI coding features.

It’s the same extension as VS Code. Install it, open a project, and let Neo handle the ML infrastructure while Cursor handles the code.

Install Extension Quickstart Guide

Installing Neo on Cursor

Cursor is built on VS Code, so Neo installs the same way through the Extensions panel.

Open Extensions

Press Cmd+Shift+X (Mac) or Ctrl+Shift+X (Windows/Linux) to open the Extensions panel in Cursor.

Search and Install

Search for “Neo” by Neo Research Inc. and click Install.

Login

Open the Neo sidebar and sign in with your account to unlock all features.

Neo + Cursor AI: How They Work Together

Cursor already comes with powerful AI features like Composer, Chat, and Cmd+K inline edits. Neo doesn’t replace any of that. Instead, it adds a layer of domain-specific AI and ML engineering that Cursor’s general-purpose AI doesn’t cover.

	Cursor AI	Neo Extension
Strength	Code generation, refactoring, inline edits	ML pipelines, data ops, experiment tracking, environment debugging
Scope	Works on the file or selection you point it at	Runs autonomous multi-step tasks across your full project
Execution	Suggests code changes for you to accept	Runs scripts, installs dependencies, and executes terminal commands directly
Integrations	Git, codebase context	AWS S3, Weights & Biases, Hugging Face, Kaggle, and more
Data	Not designed for data workflows	Load, validate, and transform datasets locally or from the cloud

Think of it this way: use Cursor AI to write the code, and use Neo to run the pipeline.

Example Workflow: Cursor + Neo Side by Side

Here’s what a typical ML development session looks like when both tools are active.

1. Write with Cursor

Use Composer or Cmd+K to scaffold a training script, write data loaders, or refactor model code.

2. Execute with Neo

Ask Neo to run the training pipeline, pull data from S3, install missing CUDA dependencies, and log metrics to W&B.

3. Auto-correct

Neo’s Auto Mode catches runtime errors, fixes environment issues, and re-runs automatically without any manual debugging.

4. Track and iterate

Review experiment logs, tweak hyperparameters in Cursor, and kick off the next run with Neo.

MCP Setup for Cursor

Neo also works as an MCP server, giving Cursor’s AI agent direct access to Neo’s tools like dataset operations, model management, and cloud integrations, right from the chat.

Open the config:

Cmd+Shift+J (Mac) or Ctrl+Shift+J (Windows/Linux), then go to Tools & MCP and select New MCP Server
Or edit ~/.cursor/mcp.json directly (restart Cursor after editing on disk)

{
  "mcpServers": {
    "neo": {
      "command": "python3",
      "args": ["-m", "neo_mcp"],
      "env": { "NEO_SECRET_KEY": "sk-v1-YOUR_KEY" }
    }
  }
}

Full MCP Documentation →

Feature Parity with VS Code

Neo’s extension is the same codebase for both editors. Everything that works in VS Code works in Cursor.

Feature	VS Code	Cursor
Extension install
Auto Mode
Sidebar and terminal
Multi-workspace support
Cloud integrations (S3, W&B, HF)
Encrypted credential vault
MCP server support