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 cli-anything-novita skill teaches an AI agent to use a command-line program that sends prompts to Novita AI, a service hosting models such as DeepSeek, GLM and MiniMax. This guide is for readers who have not used agent skills before and want the agent to call Novita models from the terminal. By the end you will know how to install the skill, check your connection, run chat requests and ask the agent to use them.

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

Quick start

1Get a Novita API key. Create an account at novita.ai and obtain an API key. Every request needs one.
2Install Python 3.10 or newer. The command-line tool requires Python 3.10+.
3Install the pip package. Run pip install cli-anything-novita to install the command-line tool.
4Save your API key. Run cli-anything-novita config set api_key with your key, or export NOVITA_API_KEY.
5Install the skill. Unzip the downloaded ZIP into ~/.claude/skills/ or a project's .claude/skills/, or upload it under Settings → Capabilities → Skills.
6Test and use. Run cli-anything-novita test, then ask the agent for a Novita-based chat or model list in plain language.

What an agent skill is and what this one does

An agent skill is a folder containing a SKILL.md file that tells an AI agent when and how to use a tool. This skill is built around the cli-anything-novita command-line tool.

Once installed, the agent can run commands to chat with a model, stream responses, list available models, test the connection and manage a saved config and chat session. The skill is designed so the agent can request machine-readable JSON output and parse it.

Chat with a model in one request
Stream a chat completion as it generates
List available models
Test API connectivity
Set, get and delete config values, and show the config file path
Check, clear or review session history

Install the skill (ZIP download or Claude apps)

Download the skill ZIP from this page and unzip it into your personal skills folder, ~/.claude/skills/, or into a project's .claude/skills/ folder if you want it limited to one project.

Alternatively, upload the skill in the Claude apps under Settings → Capabilities → Skills.

Personal: unzip into ~/.claude/skills/
Project only: unzip into the project's .claude/skills/ folder
Claude apps: Settings → Capabilities → Skills, then upload

Install the command-line tool and set your API key

The skill needs the cli-anything-novita command-line tool, which is installed from the pip package of the same name. You need Python 3.10 or newer and a Novita API key from novita.ai.

Store the key in the config file, which the source recommends, or export it as an environment variable.

Python 3.10+ is required
Get a Novita API key from novita.ai
Config file is the recommended method; NOVITA_API_KEY is the environment variable alternative

pip install cli-anything-novita
cli-anything-novita config set api_key "sk-xxx"
# or
export NOVITA_API_KEY="sk-xxx"

Run a quick connection test

Before asking the agent for anything, confirm that your key and network work. The test command takes a model ID, and models lists what is available. Use cli-anything-novita --help to see the full command list.

The default model in the source is deepseek/deepseek-v3.2. Other listed models are zai-org/glm-5 and minimax/minimax-m2.5.

test: verifies API key and connectivity
models: lists all available models
config path: shows where the config file lives

cli-anything-novita test --model deepseek/deepseek-v3.2
cli-anything-novita models
cli-anything-novita config path

Trigger the skill with natural requests

You do not need to type commands yourself. Ask the agent in plain language, naming Novita, DeepSeek, GLM or MiniMax, and the agent can choose the matching CLI commands. For example, ask it to test the Novita connection, to get a streamed answer from a model, or to list models before picking one.

If you want structured output, say so. The source says the agent should use the --json flag so its output can be parsed.

"Test my Novita AI connection using the DeepSeek model."
"Ask GLM-5 on Novita to explain quantum computing and stream the answer."
"List the models available on Novita and tell me which one suits code review."
"Use Novita with JSON output to answer: what is AI?"

Examples of commands the agent runs

These are the commands the source documents. A single chat request uses the chat subcommand with a prompt and optionally a model. The stream subcommand returns the answer incrementally. Adding --json before the subcommand gives machine-readable output.

Human-readable output is the default
Machine-readable output uses the --json flag placed before the subcommand
Calling the CLI without a subcommand opens an interactive REPL with tab-completion and history

cli-anything-novita chat --prompt "Explain quantum computing" --model deepseek/deepseek-v3.2
cli-anything-novita stream --prompt "Write a Python function to calculate factorial"
cli-anything-novita --json chat --prompt "Hello"

Safety notes: scripts, network and sensitive data

This skill runs scripts and uses the network. Every request is sent to Novita AI, a third-party hosted service, so your prompts and the model's replies leave your machine. Do not send confidential or sensitive material through it.

The source's agent guidance also asks the agent to use absolute paths for file operations and to verify that outputs exist after export operations. Check return codes (0 means success) and read stderr when a command fails.

Do not send sensitive or confidential prompts
Keep your API key in the config file or environment variable rather than in prompts
Check return codes: 0 is success, non-zero is an error
Read stderr for error messages
Use absolute paths for file operations

What people use it for

Testing whether a DeepSeek, GLM or MiniMax model answers a prompt before using it in a script
Letting an agent stream a model's answer into a terminal workflow
Listing the models available on Novita so you can pick one for a task
Having an agent parse JSON output from chat requests in a pipeline
Checking that a new API key and network connection work before a longer job

Frequently asked questions

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

Download the skill ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder. You can also upload it in the Claude apps under Settings → Capabilities → Skills. The command-line tool itself is installed with pip install cli-anything-novita and needs Python 3.10+.

What do I need to use the Novita AI skill?

You need Python 3.10 or newer, the cli-anything-novita pip package, a Novita API key from novita.ai, and internet access to the Novita API. The key can be stored with config set api_key or in the NOVITA_API_KEY environment variable.

Is cli-anything-novita free to use?

The skill is licensed under Apache-2.0. The source does not state pricing for Novita AI, so check novita.ai for its current terms, since every request requires your own Novita API key.

Which models can I use with this skill?

The source lists deepseek/deepseek-v3.2 as the default, zai-org/glm-5 from Zhipu AI and minimax/minimax-m2.5 from MiniMax. Run cli-anything-novita models to see the full list available to your account.

Why does the test command fail?

Check that your API key is set in the config file or NOVITA_API_KEY, that Python is 3.10 or newer, and that you have internet access. Read stderr for the error message and check the return code, where non-zero means an error.

Is this a good alternative to calling the Novita API directly?

The skill gives an agent a simple command-line interface with JSON output and session commands, which is useful for scripts and agents. If you need full control over API parameters or are building a web app, calling the OpenAI-compatible Novita API directly may suit you better, since the source describes Novita as OpenAI-compatible.

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 Novita API key and want an agent or script to call DeepSeek, GLM or MiniMax from a terminal with JSON output. Skip it if you have no Novita account, or if your prompts contain confidential data, because every chat request is sent to a third-party hosted service.

Download the skill