gpt2099

A Nushell scriptable MCP client with editable context threads stored in cross.stream

Features

• Consistent API Across Models: Connect to Anthropic, Cerebras, Cohere, Gemini, and OpenAI through a single, simple interface. (Add providers easily.)
• Persistent, Editable Conversations: Conversation threads are saved across sessions. Review, edit, and control your own context window — no black-box history.
• Flexible Tool Integration: Connect to MCP servers to extend functionality. gpt2099 already rivals Claude Code for local file editing, but with full provider independence and deeper flexibility.
• Document Support: Upload and reference documents (PDFs, images, text files) directly in conversations with automatic content-type detection and optional caching.

Built on cross.stream for event-driven processing, gpt2099 brings modern AI directly into your Nushell workflow — fully scriptable, fully inspectable, all in the terminal.

https://github.com/user-attachments/assets/1254aaa1-2ca2-46b5-96e8-b5e466c735bd

"lady on the track" provided by mobygratis

Getting started

Step 1.

First, install and configure cross.stream. Once set up, you'll have the full cross.stream ecosystem of tools for editing and working with your context windows.

• https://cablehead.github.io/xs/getting-started/installation/

After this step you should be able to run:

"as easy as" | .append abc123
.last abc123 | .cas

Step 2.

It really is easy from here.

overlay use -pr ./gpt

Step 3.

Initialize the cross.stream command that performs the actual LLM call. This appends the command to your event stream so later gpt invocations can use it:

gpt init

Step 4.

Enable your preferred provider. This stores the API key for later use:

gpt provider enable

Step 5.

Set up a milli alias for a lightweight model (try OpenAI's gpt-4.1-mini or Anthropic's claude-3-5-haiku-20241022):

gpt provider ptr milli --set

Step 6.

Give it a spin:

"hola" | gpt -p milli

Documentation

• Commands Reference - Complete command syntax and options
• How-To Guides - Task-oriented workflows:
  • Configure Providers - Set up AI providers and model aliases
  • Work with Documents - Register and use documents in conversations
  • Manage Conversations - Threading, bookmarking, and continuation
  • Use MCP Servers - Extend functionality with external tools
  • Generate Code Context - Create structured context from Git repositories

Reference Documentation

• Provider API - Technical specification for implementing providers
• Schemas - Complete data structure reference for all gpt2099 schemas

FAQ

• Why does the name include 2099? What else would you call the future?

Original intro

This is how the project looked, 4 hours into its inception:

https://github.com/user-attachments/assets/768cc655-a892-47cc-bf64-8b5f61c41f35