Table of content
WholeTomato

AI Agent Setup

Choose the agent you want to use with VA Intelligence and follow its section below. Each agent needs a different setup.

If Visual Assist cannot find the tool for an agent, the VA Agent window explains what is missing and shows a link and, for Codex and Claude Code, a **Locate...** option. Complete the setup for your agent, then send your message again.

Codex

Codex is the default agent. To use it, install the Codex CLI and sign in with your OpenAI account.

Install Codex CLI

  • Install the Codex CLI. If it is already installed, select "Locate Codex on this computer..." in the VA Agent window instead.
  • Send your message again.
  • If Visual Assist still cannot find the Codex CLI, restart Visual Studio to load the updated PATH, then send your message again.

Sign in to Codex

If you are not signed in, the VA Agent window shows the error "Codex is not signed in" and a Sign in with ChatGPT link.

  • Click Sign in with ChatGPT. Your browser opens the sign-in page.
  • Complete the sign-in.
  • Send your message again.

Alternatively, run codex login in a terminal, complete the sign-in, then send your message again.

Claude

Claude Code needs its CLI installed and signed in with your Anthropic account.

Install Claude CLI 

  • Install the Claude Code CLI. If it is already installed, select "Locate Claude on this computer..." in the VA Agent window instead.
  • Send your message again.
  • If Visual Assist still cannot find the Claude Code CLI, restart Visual Studio to load the updated PATH, then send your message again.

Alternatively, run claude auth login in a terminal, complete the sign-in, and then send your message again.

Ollama - gemma4 (local)

The Ollama - gemma4  agent runs on your computer through Ollama. It needs no CLI and no account, and your requests are not sent to a cloud service.

Requirements

The Ollama - gemma4 is installed in either of these ways:

  • With VAssistX (recommended).
  • Your own Ollama installation, available on PATH.

Install Ollama - gemma4 

  • Open VAssistX > Options > VA Intelligence.
  • Click Download Ollama & Gemma 4 and wait for the download to finish.
  • Return to the VA Agent window and send your message again.
  • If you already have Ollama on PATH, Visual Assist uses it and no download is needed.

If Ollama or Gemma 4 is not found, the VA Agent window shows: "Ollama and Gemma 4 were not found on this computer". The message includes:

  • A link that opens the VA Intelligence options, where you can start the download.
  • A suggestion to choose Codex or Claude Code instead.
  • After the download finishes, send your message again.