# MCP Client for Ollama (ollmcp)

A terminal application that connects local Ollama models to one or more MCP servers, supporting agent mode, multi-server setups, model switching and human-in-the-loop tool approval.

MCP Client for Ollama, or ollmcp, is a Python terminal application that lets a locally running Ollama model use tools, prompts and resources exposed by one or more MCP servers. Its agent mode runs tool calls iteratively when a model requests several in sequence, and a human-in-the-loop mode requires explicit approval before a tool call executes, with configurable defaults.

It supports managing MCP servers directly from the CLI, including scoped installs and command-line configuration, switching between installed models and their parameters, thinking-mode and reasoning-effort controls, hot-reloading a server during development, fuzzy autocomplete for commands and prompts, and saved per-provider configuration profiles. Beyond Ollama itself, it can also route to other inference providers, and its interface supports English, Simplified Chinese and Spanish.

Website: https://github.com/jonigl/mcp-client-for-ollama