Features

Every command, in detail

Five powerful AI commands, inline completions, and a chat panel — all running locally through Ollama. Here is what each one does and how to use it.

Explain Code

Select any code and get a plain-English explanation. NEURO analyzes the logic, data flow, and purpose of your selection and returns a clear, structured breakdown.

Command: NEURO: Explain Selection
Trigger: Right-click selected code, or Command Palette
NEURO AI - Explain
NEURO AI

This function calculateTotal iterates through an array of CartItem objects and computes the total cost by multiplying each item's price by its quantity, accumulating the result in a total variable.

This could be simplified using Array.reduce() for a more functional style.

Diff View - Fix Result
3
-for i in range(len(data)):
4
- result += data[i] / data[i+1]
3
+for i in range(len(data) - 1):
4
+ if data[i+1] != 0:
5
+ result += data[i] / data[i+1]
Fixed: Index out of range error, division by zero

Fix Code

Select buggy code and let NEURO detect issues and suggest fixes. The AI analyzes potential runtime errors, logic bugs, and edge cases, then provides a corrected version with explanations.

Command: NEURO: Fix Selection
Trigger: Right-click selected code, or Command Palette

Generate Tests

Automatically generate comprehensive test suites for your functions. NEURO creates tests covering happy paths, edge cases, error conditions, and boundary values.

Command: NEURO: Generate Tests
Trigger: Right-click selected code, or Command Palette
test_utils.py
1
import pytest
2
from utils import calculate_total
3
4
def test_basic_total():
5
assert calculate_total([{"price": 10, "qty": 2}]) == 20
6
7
def test_empty_list():
8
assert calculate_total([]) == 0
9
10
def test_multiple_items():
11
items = [{"price": 5, "qty": 3}, {"price": 8, "qty": 1}]
12
assert calculate_total(items) == 23
13
14
def test_zero_quantity():
15
assert calculate_total([{"price": 10, "qty": 0}]) == 0
16
17
def test_negative_price():
18
assert calculate_total([{"price": -5, "qty": 2}]) == -10
Refactor Suggestion
1
-function calc(a, b, c) {
2
- let x = a * b;
3
- let y = x + c;
4
- return y;
5
-}
1
+function calculateTotalWithTax(
2
+ price: number,
3
+ quantity: number,
4
+ tax: number
5
+): number {
6
+ return price * quantity + tax;
7
+}

Refactor

Select code to improve its structure without changing behavior. NEURO suggests better naming, cleaner patterns, type annotations, and reduced complexity.

Command: NEURO: Refactor Selection
Trigger: Right-click selected code, or Command Palette

AI Chat Panel

Open a chat panel directly in VS Code to have conversations about your code. Ask questions, get suggestions, debug issues, and explore ideas without leaving your editor.

Command: NEURO: Open Chat
Shortcut: Cmd+Shift+N (Ctrl+Shift+N on Windows/Linux)
NEURO Chat
NEURO AI mistral
You
How do I make this function async?
NEURO AI

Add the async keyword before the function and use await for the database call:

async function getUsers() {
  const users = await db.query('...');
  return users;
}

Keyboard shortcuts

Access commands quickly without touching the mouse

Command macOS Windows / Linux Context Menu
Open Chat Cmd+Shift+N Ctrl+Shift+N -
Explain Selection Command Palette Command Palette Right-click
Fix Selection Command Palette Command Palette Right-click
Generate Tests Command Palette Command Palette Right-click
Refactor Selection Command Palette Command Palette Right-click
Ask About File Command Palette Command Palette -

Configuration

Customize NEURO in VS Code settings (Cmd+,)

neuro.model

The Ollama model to use. Default: "mistral"

neuro.ollamaUrl

Ollama server URL. Default: "http://localhost:11434"

neuro.enableInlineCompletion

Enable inline code completions. Default: true

neuro.maxTokens

Maximum tokens in AI response. Default: 2048

Recommended models

Pick the model that fits your hardware and workflow

mistral
Recommended

Fast, general purpose. Great balance of speed and quality. 7B parameters, runs on 8GB RAM.

ollama pull mistral
codellama
Best for code

Optimized for code generation and understanding. Higher accuracy for coding tasks. 7B-34B sizes available.

ollama pull codellama
llama3
Latest

Meta's latest model. Excellent reasoning and instruction following. 8B and 70B sizes.

ollama pull llama3
deepseek-coder
Code specialist

Specialized code model. Strong at code completion and generation across many languages.

ollama pull deepseek-coder

Works with any language

NEURO works with any programming language your Ollama model supports

TypeScript JavaScript Python Rust Go Java C / C++ C# Ruby PHP Swift Kotlin SQL HTML/CSS Shell And more...

Architecture

How NEURO works under the hood — everything stays on your machine

VS Code
NEURO Extension
Ollama API
localhost:11434
Local Model
Mistral, CodeLlama, etc.
Response
Back to VS Code
All communication stays on localhost. No external connections.

Ready to try it?

Free, open source, and private. Download and start coding with AI in under 2 minutes.

Download NEURO AI Watch demo