Skip to content
Nx IDEDocs

Nx IDE · Nexgile Code 3.6

The AI coding agent that asks first

Nexgile Code lives beside your editor inside Nx IDE. Describe the work in plain language; it reads your codebase, plans, edits and runs commands — and stops for your approval before anything lands.

Get startedBrowse the reference
52
pages
31
modes
15+
AI providers

Documentation

Eight sections, fifty-two pages

Everything from a first install to the complete settings reference. Start anywhere — every page links to the ones around it.

Quick start

Up and running in three steps

Nexgile Code ships inside Nx IDE, so there is no second install. The only decision is which AI provider to point it at.

  1. 01

    Install and open the panel

    Nexgile Code ships inside Nx IDE — there is nothing separate to install. Open the panel beside your editor and you are ready.

    Installing & Opening
  2. 02

    Connect an AI provider

    Bring your own key: Anthropic, OpenAI, Gemini, Bedrock, Vertex, OpenRouter — or run entirely local with Ollama or LM Studio.

    Connecting a provider
  3. 03

    Run your first task

    Describe the change in plain language. Read each diff, approve what you want, and close the loop with a Task Completed summary.

    Your first task

Why teams choose it

Built so you stay in control

An agent that edits your code and runs your terminal has to be trustworthy by construction, not by good intentions.

  • You approve every consequential step

    Nothing touches your filesystem, terminal or external services until you say yes. Auto-approve is opt-in, category by category, with hard caps on per-task cost and request count acting as circuit breakers.

    Auto-approving actions

  • Free undo, always

    Every agent-initiated change is checkpointed in a shadow git repository — separate from your project's real history — so any prior state is one click away.

    Checkpoints

  • 31 modes out of the box

    Ten core modes — Architect, Code, Debug, Orchestrator, Tester, Reviewer, Shipper and more — plus 21 specialist personas. Each one is a persona with defined permissions, and you can write your own.

    Using modes

  • Bring your own provider

    Anthropic, OpenAI, Google Gemini, AWS Bedrock, GCP Vertex AI, OpenRouter and more — or run fully offline with Ollama or LM Studio. No lock-in, no markup.

    Providers overview

Providers

Bring your own model

Connect a hosted provider with your own key, or keep everything on your machine. Pick per mode — a strong model for Code and Debug, a cheaper one for Ask and docs.