claude-additional-models

claude-additional-models-mcp provides two MCP servers that adapt external AI models for Claude Desktop orchestration. The Gemini 3.x integration (gemini) exposes tools like ask_gemini, web_search, and web_reader, enabling Claude to delegate heavy reasoning and internet-backed tasks to Google Gemini via MCP while Claude handles orchestration and file I/O. The GLM-5 integration (glm5) mirrors the same tool set with the GLM-5 backend, accessible through commands such as ask_glm5 and related variants. Users add these MCP servers to Claude Desktop's configuration, enabling Claude to call the external models through a unified tool interface, thereby reducing local compute load and enabling more scalable workflows. To use, obtain API keys for the external service you want to run (Google API key for Gemini, ZAI for GLM-5), install the MCP server locally, configure Claude Desktop to point at the local MCP server, restart Claude Desktop, and verify that the tools appear in the Claude interface and respond with delegated results. The available tools provide core AI capabilities such as general analysis, summarization, and code/work product generation, with the same five-tool interface exposed for both integrations.

Prerequisites

• Node.js installed on your machine (npm included)
• Git installed
• Access credentials for external models (Google API key for Gemini; ZAI API key for GLM-5)

Install Gemini (Google Gemini 3.x)

1. Clone the repository and install Gemini dependency:

git clone https://github.com/Arkya-AI/claude-additional-models-mcp.git
cd claude-additional-models-mcp/gemini
npm install

2. Start or prepare the Gemini MCP server entry (the server exposes index.js). Ensure you have your Google API key:

GOOGLE_API_KEY=your-api-key-here node index.js

Install GLM-5 (Z.ai)

1. From the repo root, install dependencies:

git clone https://github.com/Arkya-AI/claude-additional-models-mcp.git
cd claude-additional-models-mcp
npm install

2. Start or prepare the GLM-5 MCP server entry (index.js) with your ZAI API key:

ZAI_API_KEY=your-api-key-here node index.js

Configure Claude Desktop 3) Add one or both MCP servers to Claude Desktop configuration, for example:

{
  "mcpServers": {
    "gemini": {
      "command": "node",
      "args": ["/absolute/path/to/claude-additional-models-mcp/gemini/index.js"],
      "env": {
        "GOOGLE_API_KEY": "your-api-key-here"
      }
    }
  }
}

Adjust the absolute paths to where you cloned the repo. Restart Claude Desktop after adding the servers and verify the tools list.

Verification 4) In Claude, look for the tools icon and ensure tools like ask_gemini, web_search, web_reader, parse_document, and their glm5 counterparts appear. Run a simple query to confirm that the MCP server delegates the task to the external model and returns the result to Claude.

Tips and caveats:

• Keep API keys secure; do not commit them to version control.
• The Gemini and GLM-5 MCPs expose the same core tools under different prefixes (ask_gemini vs ask_glm5). The tool behavior, prompts, and parameters are designed to be interchangeable where possible.
• Ensure network access from your host to the external model APIs; firewall rules may block outbound requests.
• If Claude Desktop reports tool discovery errors, verify the MCP server process is running, the path to index.js is correct, and environment variables are set.
• When upgrading, re-run npm install in the respective directories to ensure dependencies are up to date.