vybestack/llxprt-code

What it solves

LLxprt Code is a command-line AI coding assistant that removes the lock-in to a single AI provider. It allows developers to query and edit codebases, generate applications, and automate workflows directly from the terminal while switching between various LLM providers (including local models) seamlessly.

How it works

The tool operates as a CLI with an interactive REPL or a non-interactive mode for automation. It connects to a wide range of providers via API keys or OAuth (allowing the use of existing subscriptions like Claude Pro or ChatGPT Plus). It supports a load balancer to distribute requests across multiple accounts to avoid rate limits and can spawn specialized "subagents" with isolated contexts and different model configurations for specific tasks.

Who it’s for

Developers who prefer terminal-native workflows and want the flexibility to use multiple AI models (proprietary or open-weight) and providers without being tied to a single ecosystem.

Highlights

• Provider Agnostic: Supports Anthropic, Gemini, OpenAI, Kimi, and any OpenAI-compatible API.
• Subscription Integration: Use existing Claude Pro, ChatGPT Plus, or Kimi subscriptions via OAuth.
• Local Model Support: Works with LM Studio, llama.cpp, and Ollama for complete privacy.
• Subagents: Create isolated AI assistants with custom models and tool access for specialized tasks.
• Multi-Account Failover: Automatic failover across multiple OAuth accounts to bypass rate limits.
• IDE Integration: Native integration with the Zed editor via the Agent Communication Protocol (ACP).
• MCP Integration: Ability to connect to external tools and services via Model Context Protocol (MCP) servers.