MCP CLI client
A CLI to run LLM prompts and act as an MCP client from the terminal, with any MCP server and any model provider
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- Runs connected MCP servers and their tools
- Keeps an LLM provider key in a local config
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/mcp-cli-clientDetects the agents on your machine, checks the risk and pins the version.
Other ways to install
Assembled automatically, review before installing.
Install the tool
pipx install mcp-client-cliWithout pipx, pip install mcp-client-cli works too.
Install the client: pip install mcp-client-cli, create ~/.llm/config.json with an LLM provider and a list of MCP servers, then run prompts from the terminal.
Other ways from the author
pip install mcp-client-cliConfigure ~/.llm/config.json: the LLM provider, key and a list of MCP servers.
This is third-party code. Review the repository files before installing.
What it does
MCP CLI client is a simple command-line program that runs an LLM prompt and acts as an MCP client. From the terminal you can connect any MCP-compatible servers, so the tool serves as an alternative to Claude Desktop. The model is set in ~/.llm/config.json: provider, model name, key and, if needed, a custom base_url for OpenRouter or a local model. The same file lists MCP servers with their launch command, args, environment variables and the tools that require confirmation or are excluded.
Who it is for. For developers who want a terminal MCP client with any LLM provider.
Good fit when
- You need to call MCP servers straight from the terminal
- You want an MCP client with a provider like OpenAI, Groq or a local model
- You need an alternative to a desktop client for scripts
Not a fit when
- You want a graphical client with a UI
- You are not ready to set up a config and a provider key
Example request
Run this prompt in the terminal using the connected search MCP serverLimitations
It needs an LLM provider key in ~/.llm/config.json (or a custom base_url for a local model). Connected MCP servers run their own code and tools, and the config supports confirmation for sensitive ones. The project is small and updated infrequently.
How to disable. Uninstall the package: pip uninstall mcp-client-cli; the ~/.llm/config.json file can be removed manually.
Security check
- Runs connected MCP servers and their tools
- Keeps an LLM provider key in a local config
README in short
The README describes MCP CLI client as a simple command-line program that runs an LLM prompt and implements an MCP client. From the terminal you can connect any MCP-compatible servers, and the tool is framed as an alternative to Claude Desktop, supporting any provider including a local model via llama. Install is via pip, configured in ~/.llm/config.json: provider, model, key, an optional base_url and a list of MCP servers with commands, env and tool confirmation.
FAQ
Which providers are supported?
Any LLM provider: OpenAI, Groq, a local model via llama and others, set in the config with an optional custom base_url.
Can I connect any MCP servers?
Yes. Servers are listed in the config with a launch command, args and env, and you can mark tools that require confirmation.
Related
Open-source personal AI assistant on your own machine: answers in Telegram, Slack, Discord and WhatsApp, extended with skills and plugins
A self-improving agent from Nous Research with a TUI, messaging gateway, cron jobs and skills it writes itself
An open source coding agent for the terminal and desktop with build and plan modes
Open prompt library with a Claude Code plugin, MCP server and CLI: search, fetch and improve prompts and skills from an agent