langflow

Langflow is a platform for building and deploying AI-powered agents and workflows. The basic problem it tackles is that wiring up large language models, tools, and data sources into a working agent normally takes a lot of custom code, and that experimenting with different combinations is slow. Langflow gives you a visual authoring experience on top of that work so that you can drag and connect components on a canvas, run them, and turn the result into something other applications can call.

According to the README, Langflow comes with batteries included and supports all major LLMs, vector databases, and a growing library of AI tools. Highlight features listed include a visual builder interface for quickly getting started and iterating, source-code access in Python so you can customize any component, an interactive playground that lets you test and refine flows with step-by-step control, multi-agent orchestration with conversation management and retrieval, the option to deploy a flow as an API or export it as JSON for Python apps, and the option to deploy a flow as an MCP server so the flow becomes a tool for MCP clients. There are observability integrations with LangSmith and LangFuse, and the README also mentions enterprise-ready security and scalability. You can install Langflow as a Python package via uv (it requires Python 3.10 to 3.13), run it from source with make, or run it in Docker on port 7860. There is also a Langflow Desktop application for Windows and macOS that bundles all dependencies.

You would use Langflow when you want to prototype an agent or chain visually, share it as a working API or MCP tool, and avoid managing too much custom infrastructure. The project is open source under the MIT license, written primarily in Python, and lists topics like agents, multi-agent, large language models, and react-flow.