Client CLI MCP Server

Terminal-based LLM interface for executing prompts and MCP server connections.

Data last scanned yesterday · Reviewed 3 weeks ago

Overview

Client CLI MCP Server provides a terminal interface to interact with MCP-compatible servers and various LLM providers, including OpenAI, Groq, and local models via llama.cpp. It serves as an alternative to existing desktop clients by enabling prompt execution directly from the command line. Users define their infrastructure through a configuration file, allowing for granular control over enabled servers, system prompts, and tool execution. The client supports piping text input and image data, facilitating integration into shell-based workflows and automation tasks.

Our verdict

This tool carries a high risk level with a security score of 39/100. It is a community-maintained project that is not editorially reviewed and lacks an official maintainer. Most notably, the tool defaults to full write access without a restricted read-only mode, and the lack of OAuth support necessitates careful management of API keys within its local configuration files. Use only in trusted development environments.

  • Supports local LLM execution via llama.cpp integration
  • Zero direct dependencies minimizes supply-chain surface area
  • Configurable tool execution confirmation requirements
  • No read-only mode creates inherent write access risks
  • Lacks OAuth support for secure credential handling
  • Last code update occurred over 250 days ago

Compatibility

ClientLocalDockerRemoteRead-only
ChatGPT
Claude Desktop
Cursor
VS Code
Windsurf

Frequently asked questions

Where can I store the configuration file?

You can place your config.json file in either ~/.llm/config.json or $PWD/.llm/config.json.

Can I use environment variables instead of hardcoding API keys in the config file?

Yes, you can provide your LLM API key via the LLM_API_KEY or OPENAI_API_KEY environment variables.

How can I prevent the client from asking for confirmation before running a tool?

You can use the --no-confirmations flag when running the command to bypass user confirmation requirements for tools.

Can I exclude specific tools provided by an MCP server?

Yes, you can specify a list of tool names in the exclude_tools configuration field within your config.json file.

Are there specific requirements for using the CLI in automated scripts?

Yes, you should use the --no-intermediates flag to suppress intermediate output and only return the final result.

Does the configuration file support comments?

Yes, you can include // comments within the JSON configuration file to document your settings or switch configurations easily.

Changelog

Badge

Client CLI MCP Server security score, rated on RepoAI

Maintain this server? Add the live badge to your README.