Setup guides
Turn on Tally’s XML server once, then add MCP for Tally to your AI app. The app does the second part with one click; these guides show what it does.
Start with Tally
Then your AI app
- Claude DesktopClaude Desktop for Windows and Mac starts MCP for Tally itself and talks to it privately on your computer.
- ChatGPT DesktopChatGPT Desktop for Windows can start a local MCP server. chatgpt.com in a browser and the phone apps cannot reach a program on your computer.
- CursorCursor reads MCP servers from mcp.json in your home folder.
- VS CodeVS Code (GitHub Copilot in agent mode) reads MCP servers from your user mcp.json.
- Claude CodeClaude Code adds MCP servers with one command.
- Ollama (fully local)Ollama runs AI models on your own computer. With AnythingLLM as the chat window, your question, your books and the answer never leave the PC.
- LM StudioLM Studio runs AI models on your own computer, so nothing about your books leaves it at all.
- WindsurfWindsurf reads MCP servers from its mcp_config.json.
- Gemini CLIGemini CLI reads MCP servers from its settings.json.
- Codex CLIOpenAI’s Codex CLI reads MCP servers from config.toml.
- Any other MCP appAny app that can start a local MCP server, or connect to one by address, can use MCP for Tally.