MiniMax Code

MiniMax Code is a desktop AI agent from MiniMax that combines a coding assistant with a broader local workspace for research, file handling, browser tasks, and automation. Rather than only suggesting code completions, it is built to accept a goal, break it into steps, use local tools, and continue working through longer tasks with liMITed supervision.

What it does

The product brings chat, project workspaces, File Operations, a terminal, a built-in browser, reusable Skills, memory, and scheduled tasks into one environment. Users can describe a development task in natural language, and the Agent can inspect a project, edit files, run commands, review changes, and iterate. A separate Work mode is AImed at General knowledge work, where the focus is on progress and deliverables rather than low-level technical details.

Key capabilities

• multi-agent task handling: A lead agent can split complex work into smaller assignments, coordinate parallel workers, and use verification steps to catch errors.
• Long-running goals: Goal mode is designed for peRSIstent objectives that may require multiple rounds of planning and execution.
• Local project context: The desktop app can bind to local workspaces and work with files, terminal commands, and code diffs.
• MEMOry and skills: A layered memory system and reusable skills help the agent retain context and repeat established procedures.
• Browser and Automation: Built-in browser control, scheduled tasks, and remote control options extend the agent beyond pure coding.
• Model flexibility: MiniMax models are the primary option, and bring-your-own-key support is described for compatible endpoints.

Who it is for

MiniMax Code is aimed at developers who want an agentic assistant for implementation, debugging, and project work, as well as technical users who need help with research, Office Automation, and recurring Workflows. Teams evaluating AI coding agents may find the multi-agent and verification APProach useful, while indiVidual builders may prefer the desktop workspace for local projects.

How to get started

1. Download the desktop CLIent for Windows or macOS, or use the Web option if available.
2. Open or create a project workspace and connect the local files the agent should access.
3. Choose Coding or Work mode depending on whether the task is software development or general delivery work.
4. Describe the goal, review proposed ACTions, and approve sensitive commands or browser Operations.

Pricing and access caveat

Access is described through a token-based subscription plan, prepaid credits, and limited free daily credits. Exact pricing, plan limits, model availability, and regional access may vary, so check the current product pages before relying on a specific tier. Users should also review permission settings for terminal commands, file changes, and browser actions, especially when working with sensitive local projects.