Getting started

Use Claude Pro or Max for Roblox Studio with BloxBot

November 5, 2026

Short version: BloxBot 0.14.0 adds a provider called Claude Agent. It runs the Claude Code already signed in on your computer, through Anthropic's Agent SDK, so the Roblox Studio work you do in BloxBot draws on your Claude Pro or Max plan instead of pay-as-you-go API credits. BloxBot shows a clear warning first: this is at your own risk.

Until now, Claude in BloxBot meant paying per token. Claude plan logins are meant for Anthropic's own apps, so a desktop app can't just ask for your Claude password. Claude Agent takes a different route. It doesn't log in to Claude at all. It starts Claude Code, Anthropic's own tool, and Claude Code uses the login it already has. BloxBot gets this from OpenCode, the open-source engine BloxBot runs on, which ships the claude-agent provider in its own desktop app.

How to connect your Claude plan

Update to BloxBot 0.14.0 or later. New to BloxBot? The setup guide covers the first run and the Studio connection.
Install Claude Code by following Anthropic's setup page.
In a terminal, run claude auth login and sign in with the Claude account that has your plan.
In BloxBot, open Settings → Providers and choose Claude Agent (“Use your Claude plan or another Claude Code login”).
Read the warning, tick I understand and want to continue, and click Connect. BloxBot checks that Claude Code is installed and signed in, then lists the models it can use.
Pick a Claude model in the message box and start a chat.

On a Mac, BloxBot now reads your shell's PATH the way OpenCode's desktop app does, so it finds claude even when you open BloxBot from the Finder. It passes on only PATH, not your other environment variables, because those often hold credentials.

What are the risks of using your Claude plan?

The warning in the app reads: “BloxBot isn't affiliated with Anthropic. Its terms may limit using Claude plans in other apps, and anything that happens to your account is your responsibility.” That is the honest position. Anthropic decides how a Claude plan may be used, and it can change that. Read Anthropic's legal and compliance page before you connect.

If you would rather not take that risk, you have three routes with no plan-terms question:

An Anthropic API key. Pay per token, no plan rules to think about.
Your ChatGPT plan. OpenCode supports signing in with ChatGPT Plus or Pro. See using your ChatGPT or Codex subscription.
Free or cheap open models. The free models BloxBot includes, or OpenCode Go from $10 a month.

What BloxBot does and doesn't touch

Your login stays with Claude Code. BloxBot never sees the token. It only checks that Claude Code answers.
Roblox guidance goes in BloxBot. 0.14.0 stops BloxBot from loading skills or instructions from ~/.claude, ~/.agents or ~/.opencode. Put Roblox-specific guidance in Settings → Instructions. Claude Code itself may still read its own user settings when it runs, so keep anything there general.
Session files stay where Claude Code expects them. Claude Code still writes its session history to ~/.claude/projects. Moving it would sign you out.

Is a Claude plan good for Roblox work?

Claude models have led Roblox's own OpenGameEval benchmark, with Claude Fable 5 first on Pass@1 for both code generation and debugging. A plan also makes long agentic sessions predictable: you stop at a usage limit instead of running up a bill. The trade-off is the same as with a ChatGPT plan. An agent that inspects your place through Studio's MCP spends tokens on every look around, so heavy days hit the limit sooner than ordinary chat does. Name the object you want changed and say how you will check the result; our notes on bounded tasks show why that helps. Comparing options? See how BloxBot compares to other Roblox AI tools.

Frequently asked questions

Can I use my Claude Pro or Max subscription in BloxBot?

Yes, from BloxBot 0.14.0. Install Claude Code, sign in with claude auth login, then connect the Claude Agent provider in BloxBot's Settings, Providers. BloxBot runs your own Claude Code through Anthropic's Agent SDK, so the work counts against your Claude plan instead of API credits.

Why does BloxBot say Claude Agent is at my own risk?

BloxBot isn't affiliated with Anthropic, and Anthropic's terms may limit how a Claude plan can be used in other apps. BloxBot shows that warning and keeps Connect disabled until you tick that you understand. If you want no ambiguity, use an Anthropic API key instead.

Does BloxBot see my Claude login?

No. Claude Agent uses the login Claude Code already has on your computer. BloxBot never receives the token; it only checks that Claude Code is installed and signed in.

Which Claude models can I use through Claude Agent?

The ones your Claude Code login can use. BloxBot lists what Claude Code reports, so a Max plan typically shows more than Pro.

Does Claude Agent work on Windows?

It needs Claude Code installed and on your PATH. The BloxBot team tested it on macOS for the 0.14.0 release; Windows was not part of that test pass yet.

Sources

Back to the guide