Guide · Agent skill

How to use the Novita AI command-line client agent skill

Updated 2026-10-29 · written with AI from the project's own documentation and checked against our catalogue data

The Novita AI command-line client agent skill teaches an AI agent to use a command-line tool that talks to Novita AI, a service hosting open models such as DeepSeek and GLM. Once installed, the agent can send prompts, stream answers, list models and test that your API key works by running commands instead of writing its own API code. This guide explains what the skill does, how to install and trigger it, and what to watch for before you send data to a third-party provider.

Novita AI command-line clientHKUDS/CLI-Anything · Apache-2.0 Download the skill

Quick start

1Check prerequisites. Confirm you have Python 3.10 or newer and a Novita API key from novita.ai.
2Install the package. Run pip install cli-anything-novita in your terminal.
3Download and unzip the skill. Download the skill ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder.
4Or upload the skill in Claude. In the Claude apps, open Settings → Capabilities → Skills and upload the ZIP instead.
5Save your API key. Run cli-anything-novita config set api_key with your key, or export NOVITA_API_KEY.
6Test the connection. Run cli-anything-novita test --model deepseek/deepseek-v3.2 to verify the key and connectivity.
7Ask the agent in plain language. Make a natural request such as comparing models; the agent will run the matching command, using --json where it needs structured output.

What is an agent skill and what does this one do?

An agent skill is a folder containing a SKILL.md file that gives an AI agent instructions for a specific task. This skill, named cli-anything-novita, describes a command-line interface for Novita AI, an OpenAI-compatible API service for models like DeepSeek, GLM and others.

With the skill in place, the agent can run commands such as chat, stream, models and test, and can request machine-readable JSON output so it can read the results.

Install the cli-anything-novita package

The CLI is installed as part of the cli-anything-novita Python package. You need Python 3.10 or newer and a Novita API key from novita.ai.

Confirm Python 3.10+ is available on your machine.
Get a Novita API key from novita.ai.
Install the package with pip.

pip install cli-anything-novita

Install the skill for Claude

Download the skill ZIP from this page and unzip it into one of these locations:

~/.claude/skills/ for personal use across projects
A project's .claude/skills/ folder for use in one project
Or, in the Claude apps, go to Settings → Capabilities → Skills and upload the ZIP.

Set your API key

The CLI needs your Novita API key. The source recommends saving it to the config file, though you can also use an environment variable.

Config file (recommended): set the key with the config command.
Environment variable: export NOVITA_API_KEY in your shell.

cli-anything-novita config set api_key "sk-xxx"
export NOVITA_API_KEY="sk-xxx"

Trigger the skill with natural requests

You do not need to type commands yourself. Ask the agent in plain language, for example to compare how DeepSeek and GLM answer the same question, or to check whether your Novita key works. The agent should then choose the matching cli-anything-novita command.

Useful commands the agent can run include test to check connectivity, models to list models, chat for a single answer and stream for streamed output. Add --json before the subcommand to get structured output.

Check my Novita key works with the test command.
List the models available on Novita.
Ask DeepSeek and GLM the same question and compare the answers.
Stream a Python factorial function from deepseek.

cli-anything-novita test --model deepseek/deepseek-v3.2
cli-anything-novita models
cli-anything-novita --json chat --prompt "Hello"

Models and output formats

The source lists three default models. deepseek/deepseek-v3.2 is the default. The output is human-readable by default, with tables and formatted text, and the --json flag switches to structured output for agents and scripts.

When the agent uses this CLI programmatically, the source advises it to always use --json, check return codes (0 for success, non-zero for errors), parse stderr on failure, use absolute paths for file operations and verify outputs exist after exports.

deepseek/deepseek-v3.2 (DeepSeek, default)
zai-org/glm-5 (Zhipu AI)
minimax/minimax-m2.5 (MiniMax)

Safety notes before you use it

This skill runs scripts and uses the network. Requests go to Novita's servers, so prompts and any data in them leave your machine and are sent to a third-party AI provider. Do not paste sensitive or confidential data into prompts.

Nothing works without a Novita account and API key. The skill does not run a local model, so it is not an offline tool. Review the package before installing if you need to confirm what runs on your machine.

Avoid confidential data in prompts.
Keep your API key out of shared files; the config file or environment variable are the documented options.
Expect network use on every chat, stream or test call.

What people use it for

Querying DeepSeek, GLM or MiniMax from the terminal and comparing their answers side by side.
Letting an agent list the models Novita offers before choosing one for a task.
Building scripts that call Novita-hosted models and need JSON output to parse.
Checking that an API key and connection work before running a larger job.

Frequently asked questions

How do I install the Novita AI command-line client skill?

Install the Python package with pip install cli-anything-novita, then download the skill ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder. In the Claude apps you can instead upload the ZIP under Settings → Capabilities → Skills.

What do I need to use it?

You need Python 3.10 or newer, the cli-anything-novita pip package, and a Novita AI API key from novita.ai. Without an API key nothing works, and the skill uses the network.

Is the Novita AI skill free?

The source does not state pricing or licence terms for Novita AI usage. The repository containing this skill is listed under Apache-2.0, but check Novita's own site for API costs.

Can I run a local model with this skill?

No. Requests go to Novita's servers, so the skill does not run a model on your machine.

How do I fix a failing connection or bad API key?

Run cli-anything-novita test --model deepseek/deepseek-v3.2 to verify your API key and connectivity. Check that the key is saved with cli-anything-novita config set api_key or in the NOVITA_API_KEY environment variable.

How is this different from calling the Novita API directly?

The skill lets the agent use ready-made commands instead of writing its own API code. It also offers a --json output mode designed for agent consumption.

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

Use this skill if you have a Novita account and want an agent to query DeepSeek, GLM or MiniMax from the terminal, or to feed JSON results into scripts. Pick something else if you need offline or local models, or if you plan to send confidential data to a third-party provider.

Download the skill