Skip to content
Get packs

Guide · Agent skill

How to use the MiniMax AI Command-Line Tool agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

This guide explains the cli-anything-minimax agent skill, a command-line tool that lets an AI agent talk to MiniMax AI models and turn text into speech files. It is aimed at people who may never have used an agent skill before. By the end you will know how to install the skill, make it run chat and text-to-speech requests from plain-language prompts, and what safety points to check first.

MiniMax AI Command-Line ToolHKUDS/CLI-Anything · Apache-2.0

Quick start

  1. 1Check Python. Confirm you have Python 3.10 or newer installed before you start.
  2. 2Get an API key. Create or sign in to your account at platform.minimax.io and copy your MiniMax API key.
  3. 3Install the command-line tool. Run the pip install command from the HKUDS/CLI-Anything repository for the minimax agent harness.
  4. 4Add the skill files. Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills in the Claude apps.
  5. 5Set your API key. Export MINIMAX_API_KEY or run config set api_key with your key.
  6. 6Test the connection. Run cli-anything-minimax test to confirm the tool can reach the MiniMax service.
  7. 7Ask for a result. Describe the chat answer or speech file you want in plain language, and the agent will run the matching command.

What does the cli-anything-minimax skill let an agent do?

The skill teaches an AI agent to use a command-line program called cli-anything-minimax, which connects to the MiniMax AI service. Once it is set up, the agent can send chat prompts to MiniMax models, stream responses, list available voices, and generate speech audio from text, saving the result to a file.

The tool requires your own MiniMax API key. Prompts and text are sent to the MiniMax online service, so this skill is not suited to offline or private processing.

  • Chat with MiniMax-M3 (the default) or MiniMax-M2.7
  • Stream a chat response as it is generated
  • List available voice IDs and TTS models
  • Turn text into speech with options for speed, volume, pitch, sample rate, bitrate, format and channels
  • Return JSON output for scripts and agent workflows

How do you install the agent skill?

There are two parts: the skill files for your agent, and the command-line tool itself. Install the skill files first. Download the ZIP from this page and unzip it into ~/.claude/skills/ for all your projects, or into a project's .claude/skills/ folder for one project only. If you use the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

Next, install the command-line tool. The source gives this pip command, which installs it from the HKUDS/CLI-Anything GitHub repository. You need Python 3.10 or newer.

  • Requires Python 3.10 or newer
  • Requires a MiniMax API key from platform.minimax.io
  • The skill package contains a single file, SKILL.md
pip install git+https://github.com/HKUDS/CLI-Anything.git#subdirectory=minimax/agent-harness

How do you add your MiniMax API key?

The tool cannot work without a key. You can set it as an environment variable for the current shell session, or save it in the tool's configuration with the config command. The source does not say where the saved value is stored, but the config path command shows the location of the config file.

Keep your key private. Anyone with the key can use your MiniMax account.

  • Environment variable: export MINIMAX_API_KEY followed by your key
  • Saved setting: config set api_key with your key
  • Check the config file location with config path
export MINIMAX_API_KEY="your-api-key" # or cli-anything-minimax config set api_key "your-api-key"

How do you ask the agent to use it?

You do not need to type commands yourself. Once the skill is installed, describe the result you want in plain language, such as asking for a short answer from MiniMax or for an MP3 narration of a paragraph. The agent then picks the matching command.

If you want to run the tool directly, these are the main commands from the source. Add --help to any command to see its options.

  • Chat: a single prompt that returns one answer
  • Stream: a prompt that returns the answer progressively
  • TTS: text in, audio file out, with --output naming the file
  • Voices: lists voice IDs you can pass to --voice
  • Test: checks API connectivity, useful when something fails
cli-anything-minimax --help cli-anything-minimax chat --prompt "Explain quantum computing" cli-anything-minimax stream --prompt "Write a Python quicksort" cli-anything-minimax tts --text "Hello!" --output hello.mp3 cli-anything-minimax tts --text "Fast" --model speech-2.8-turbo --voice English_Insightful_Speaker --output fast.mp3 cli-anything-minimax --json chat --prompt "Hello"

What TTS options and models are available?

Text-to-speech takes --text as required input. You can also set the model, the voice, the output file and several audio controls. The source gives these ranges and choices; run tts --help to see the defaults.

The source lists several TTS models. speech-2.8-hd is the default high-definition model, and the turbo models are described as fast. The chat models are MiniMax-M3 (default) and MiniMax-M2.7.

  • --speed from 0.5 to 2.0
  • --vol from 0 to 10
  • --pitch from -12 to 12
  • --sample-rate: 8k, 16k, 22.05k, 24k, 32k or 44.1k
  • --bitrate: 32k, 64k, 128k or 256k
  • --format: mp3, pcm or flac
  • --channel: 1 or 2
  • TTS models: speech-2.8-hd, speech-2.8-turbo, speech-2.6-hd, speech-2.6-turbo, speech-02-hd, speech-02-turbo, speech-01-hd, speech-01-turbo
cli-anything-minimax models cli-anything-minimax models --tts

What should you check for safety and network use?

The skill runs scripts and uses the network. Every chat prompt and every piece of text sent to speech is transmitted to the MiniMax online service. Do not send confidential or personal material unless you are comfortable with that.

The source's guidance for agents is useful for you too. Check return codes, where 0 means success and any non-zero value means an error. On failure, read the error messages on stderr. For speech output, use absolute file paths so you know exactly where the audio was saved.

  • Prompts and text leave your machine and go to MiniMax
  • Your API key grants access to your MiniMax account, so store it carefully
  • Use --json when a script needs to parse the output
  • Use absolute paths for TTS output files
  • The skill is licensed under Apache-2.0

Tips for getting reliable results

Start with the test command after installing, so you know the key and connection work before relying on the skill. Use the --json flag whenever another program will read the output. If a command fails, run it again with --help to confirm the options you used are valid.

The session commands show status, clear history, and list past commands, which helps when an agent's earlier requests affect later ones.

  • Run test first to confirm connectivity
  • Use --json for parseable output
  • Use session status, session history and session clear to manage session state
  • Use config get to check which settings are saved
cli-anything-minimax test cli-anything-minimax --json chat --prompt "Hello"

What people use it for

  • Generate a spoken narration of a script as an MP3 file using a chosen MiniMax voice
  • Ask MiniMax-M3 a question from a terminal and get the answer as JSON for a script to parse
  • Stream a long chat response while working in the terminal
  • List available voice IDs before choosing one for a speech project
  • Build an agent workflow that sends text to MiniMax and saves audio at an absolute file path

Frequently asked questions

How do I install the cli-anything-minimax skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills. The command-line tool itself is installed with pip from the HKUDS/CLI-Anything GitHub repository.

What do I need to run cli-anything-minimax?

You need Python 3.10 or newer and a MiniMax API key from platform.minimax.io. The tool cannot function without a key. It also needs network access because prompts and text are sent to the MiniMax online service.

Is cli-anything-minimax free to use?

The source is licensed under Apache-2.0, but it does not state what MiniMax charges for API access. Check platform.minimax.io for current pricing before you rely on it.

Can I use it offline or with private data?

No. The tool sends chat prompts and speech text to the MiniMax online service, so it is not suitable for offline or confidential processing.

What should I do if a command fails?

Run cli-anything-minimax test to check connectivity and confirm your API key is set. Check the return code, where 0 means success and anything else means an error, and read the error message on stderr.

How does it compare with other speech or chat tools?

The source does not compare cli-anything-minimax with other tools. It covers only MiniMax chat and text-to-speech, so it is not a general-purpose agent skill.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Choose this skill if you already have a MiniMax account and want your agent or scripts to call MiniMax chat and text-to-speech from the terminal. Pick something else if you do not have a MiniMax API key, need offline or private processing, or want a general-purpose skill that covers more than MiniMax.

Download the skill