Read: From using AI to building AI systems, a defining note on what I’m exploring.

LangREPL

An interactive command-line REPL for building and running multi-tool, multi-model agents powered by LangChain and LangGraph.

midodimori · Since 2025-10-25
Loading score...

Overview

LangREPL is an interactive terminal chat application built with LangChain, LangGraph, Prompt Toolkit and Rich. It enables fast experimentation and debugging of multi-tool, multi-model agents via a REPL or server mode with LangGraph Studio integration.

Key features

  • Interactive REPL and server mode (LangGraph Studio support for visual debugging).
  • Persistent conversation storage, todo/approval workflows, and subagent support for complex tasks.
  • Extensible tool system and multi-provider LLM support for production-like experimentation.

Use cases

  • Research agents that gather and synthesize information into structured outputs.
  • Developer workflows for validating agent configurations, tools and model behavior.
  • Automation pipelines where tasks are decomposed into tool-driven subtasks.

Technical highlights

  • Native Python implementation with pip/poetry/uv packaging for easy integration.
  • Modular architecture and example-driven configs facilitate extensibility and deployment.
  • MIT licensed and community-maintained for reuse and contribution.

Comments

LangREPL
Score Breakdown
💻 CLI 🛠️ Dev Tools