Skip to content
AgoraMod Launcher

Connect

Get help from an AI assistant

Let an AI assistant app on your computer read your crash logs and fix problems through Agora, with your approval.

Two versions of this page follow. Step by step walks through the task in order. In depth explains the underlying behavior. They describe the same product, at different levels of detail.

Step by step

Follow the steps in order. No prior modding knowledge assumed.

An AI assistant app on your computer can use Agora's command-line tool to look at your instances, read crash reports, and suggest fixes. It asks before it changes anything, and you approve each step. Crash Doctor is still the first place to look: this is for when you want help working through a problem.

After this you will be able to

  • Pick an assistant app that suits you.
  • Hand the assistant Agora's command-line tool with one copied message.
  • Keep control of what the assistant changes.

Pick an assistant app

You need an assistant app that can run commands on your computer. A chat window in a web browser cannot. Here are a few suggestions to get you started, if you already have a premium account with an AI company, research to see if they include a "coding agent" first before picking one of these.

  • OpenCode desktop: an open source and quick to set up AI app with a few free models. It can also use accounts from several other AI companies.
  • Codex in the ChatGPT app: OpenAI's assistant, included with a ChatGPT account. How much you can use it depends on your plan.
  • Claude Code: Anthropic's assistant, in the Claude desktop app. It needs a paid Claude plan, so it suits you if you already have one.
  • Google Antigravity: Google's assistant app. Sign in with a Google account.

Ask for help

Agora comes with a command-line tool the assistant uses to work with your instances. Agora gives you a ready-made message that tells the assistant where that tool is and how to use it safely.

  1. In Agora, open Settings, then Services, then AI assistant, and choose Copy message.
  2. Open your assistant app and paste the message, then finish it with your problem, such as "my Survival instance crashes on launch".
  3. The assistant asks to run commands. Reading commands, such as listing instances or reading a crash report, are safe to allow.
  4. Before it changes anything, it should explain what and why. Allow the change only if it makes sense to you.
  5. When it asks you to test, launch the instance from the Agora app as usual and tell it what happened.

What the assistant sees

Anything the assistant reads is sent to the company that runs it. Crash reports can contain your computer's username, folder names, server addresses, and chat messages. If that matters to you, use Crash Doctor's redacted report instead.

In depth

How the system actually behaves, and why, for readers who prefer the model to a recipe.

How the assistant learns to use Agora, what it can reach on your computer, and when the MCP route suits you better. The approval prompts in your assistant app are what keep you in control.

After this you will be able to

  • Understand what Agora skill gives an agent.
  • Know where the trust boundary really is.
  • Know when MCP is the better choice.

What Agora skill provides

The agora skill command prints out a markdown format guide that teaches the agent how to use Agora's command-line tool and gives a few recommended practices.

  • The agent reads results with --json and checks exit codes.
  • The CLI and desktop app share the same folder and the desktop app is configured to reload data changed by the CLI, so you can run both simultaneously with ease.
  • The guide asks the agent to have you launch from the app for your own convenience and security, you can simply ask the agent to launch for you if you want!

Where the trust boundary is

An agent with a terminal can do anything your account can, including editing instance files without Agora. Since AI sometimes makes mistakes, Agora's important commands stop and tell the agent to ask you before continuing. We recommend keeping the app's per-command approval on, locking instances you care about, and pointing experiments at a disposable directory.

When to use MCP instead

MCP suits AI apps that cannot run commands but can connect to a local tool server. There, Agora's MCP approval queue is the only way in, so it does protect you. It takes more setup: a token, a config file, and the Agora app running. See the MCP and external AI tools guide.

This guide is the same text Agora shows in its built-in Help & Guide, which is searchable and links directly to the screens it describes.