gpt.nu

gpt.nu - LLM Integration for Nushell

gpt.nu is a Nushell module that enables direct interaction with large language models (LLMs) from your command line. It provides a simple interface supporting multiple LLM providers and the Model Context Protocol (MCP).

With gpt.nu, you can:

• Chat with AI models using one consistent API
• Maintain conversation context across sessions. In addition to storing the conversation, cross.stream makes it convenient to review and manually edit conversation threads, giving you direct control over the context window.
• Integrate with MCP servers to extend tool capabilities
• Receive streamed AI responses directly in your shell

Built on cross.stream for event processing, gpt.nu brings modern AI capabilities directly into your Nushell workflow, without leaving the terminal.

Usage

First install and configure cross.stream: https://cablehead.github.io/xs/getting-started/installation/

use gpt

gpt init

Installs the required dependencies into your cross.stream store, and prompts you to select an LLM provider, an API key, and a model.

gpt

Makes a request to the LLM provider.

gpt thread

Review the current conversation thread.

gpt mcp

Model Context Protocol (MCP) servers are really just CLI tools that read from stdin and write to stdout.

cross.stream generators spawn CLI tools and package each line of output into event frames (.recv) while routing frames ending in .send as input, turning them into services you can interactively poke at.

gpt mcp leverages this approach to provide a hands-on environment for experimenting with and understanding MCP servers.

Features

• Spawn an MCP server as a cross.stream generator
• List available tools on the server.

Spawn an MCP Server

Register your MCP server:

# gpt mcp register <name> <command>
gpt mcp register filesystem 'npx -y "@modelcontextprotocol/server-filesystem" "/project/path"'

List Available Tools

List the tools provided by the MCP server:

gpt mcp tools list filesystem

──#──┬───────────name────────────┬─────────────────────────────────────────────────────────────────────────description─────────────────────────────────────────────────────────────────────────┬─...─
 0   │ read_file                 │ Read the complete contents of a file from the file system. Handles various text encodings and provides detailed error messages if the file cannot be read.  │ ...
     │                           │ Use this tool when you need to examine the contents of a single file. Only works within allowed directories.                                                │
 1   │ read_multiple_files       │ Read the contents of multiple files simultaneously. This is more efficient than reading files one by one when you need to analyze or compare multiple files │ ...
     │                           │ . Each file's content is returned with its path as a reference. Failed reads for individual files won't stop the entire operation. Only works within allowe │
     │                           │ d directories.
 ...