MCPHost

MCPHost is an MCP host and CLI application that enables Large Language Models to interact with external tools through the Model Context Protocol. It acts as a runtime bridge connecting supported language models, including Anthropic Claude, OpenAI, Google Gemini, and local Ollama instances, to one or more MCP servers. Developers, automation engineers, and command-line users employ MCPHost to give AI models direct access to external utilities, data stores, and scripts. The application handles tool discovery, execution pipelines, and context management across sessions. It supports both interactive conversational terminal sessions and non-interactive scripted workflows using YAML automation scripts. MCPHost also offers an embedded Go SDK, allowing programmers to execute MCP-based agent prompts within their own Go binaries without managing external child processes. Through granular tool filtering, developers can restrict or grant specific capabilities per server. Note that active development has officially stopped, and the project is now archived in favor of Kit.

Category: AI & LLM Tooling

Tags: cli, integration, llm, protocol

Visit MCPHost

How to install and configure MCPHost

  1. Ensure you have Go 1.23 or later installed on your system. 2. Install MCPHost globally using Go: bash go install github.com/mark3labs/mcphost@latest 3. Configure your API key by exporting environment variables such as OPENAI_API_KEY, ANTHROPIC_API_KEY, or GOOGLE_API_KEY. 4. Create or edit the configuration file at ~/.mcphost.json or ~/.mcphost.yml to define your local MCP servers under mcpServers. 5. Launch the host in interactive mode by running: bash mcphost For SDK integration details and advanced options, refer to the MCPHost repository. Note that the project is archived in favor of Kit.

What you can do with MCPHost

  • Running interactive command-line terminal conversations with Claude or OpenAI connected to local filesystem tools. - Automating routine administration tasks using non-interactive execution and YAML-based automation scripts. - Embedding Model Context Protocol capabilities directly into Go applications using the provided native SDK. - Interfacing local Ollama-hosted open-source language models with external MCP tool servers securely. - Restricting model access to specific dangerous tools using granular allowedTools and excludedTools configuration rules.

Key facts

  • Open Source
  • https://github.com/mark3labs/mcphost
  • AI & LLM Tooling
  • cli, integration, llm, protocol

Part of MCP Servers

Related MCP servers

  • MCP Memory Dashboard โ€” MCP Memory Dashboard is an MCP server desktop interface that connects to the MCP Memory Service to provide visual semanticโ€ฆ
  • MCP Manager โ€” MCP Manager is an MCP server management tool that connects directly to your Claude Desktop environment, enabling users to discover,โ€ฆ
  • MCP Lab โ€” MCP Lab is an MCP server development environment designed for building, testing, and debugging custom Model Context Protocol servers integratedโ€ฆ
  • MCP LLM Integration Server โ€” MCP LLM Integration Server is an MCP server that connects local Large Language Model runtimes with Model Context Protocol clientsโ€ฆ
  • MCP Neurolora โ€” MCP Neurolora is an MCP server that provides code analysis, code collection, and automated documentation generation using the OpenAI API.โ€ฆ
  • MCP OpenVision โ€” MCP OpenVision is an MCP server that provides image analysis capabilities powered by OpenRouter vision models. It connects client interfacesโ€ฆ

How do I install MCPHost?

You can install MCPHost by running go install github.com/mark3labs/mcphost@latest with Go 1.23 or later installed. After installation, configure your provider API keys and define your servers in ~/.mcphost.json or ~/.mcphost.yml.

What can MCPHost do?

MCPHost serves as a CLI host application connecting LLMs from Anthropic, OpenAI, Google, or Ollama to external tools via the Model Context Protocol. It supports interactive chat, headless script execution, tool filtering, and a native Go SDK for custom applications.

Which LLM providers work with MCPHost?

MCPHost supports Anthropic Claude models, OpenAI models, Google Gemini models, and any local Ollama models that provide function-calling support. It also connects to any API endpoint compatible with the OpenAI API format.

Is MCPHost still actively maintained?

No, active development on MCPHost has stopped and the repository has been archived. The creators recommend that users migrate to Kit, a successor project that builds on MCPHost with an updated and extensible architecture.

Is MCPHost open source?

Yes, MCPHost is an open-source project hosted on GitHub by mark3labs. Users can inspect the Go source code, build it locally, or use its SDK components under its open-source license.

  • AI Tools
  • Categories
  • Industries
  • CLI Coding Agents
  • MCP Servers
  • MCP Categories