langflow-ai/langflow
Langflow is an open-source development platform created for software developers building and deploying artificial intelligence agents and multi-agent workflows. It couples an interactive visual builder with direct Python source code access, enabling custom component programming alongside a step-by-step playground for testing execution paths. Core platform capabilities include multi-agent orchestration, conversation management, contextual retrieval, and built-in observability integrations with LangSmith and LangFuse. Flows can be exported as JSON for Python applications, served directly as standalone REST APIs, or hosted as Model Context Protocol (MCP) servers that transform workflows into callable tools for external MCP clients. Additionally, standalone desktop installers are supplied for macOS and Windows systems with all dependencies bundled. The GitHub snapshot records an MIT license.
Local execution requires Python 3.10 through 3.14 alongside the recommended uv package manager. Developers install the software using uv pip install langflow -U. Because official release packages disable automatic login by default, users must configure the LANGFLOW_SUPERUSER_PASSWORD environment variable before launching the server, then sign in with the default username langflow. The application starts via uv run langflow run at http://127.0.0.1:7860. Alternative execution methods include building from source with make run_cli or deploying via Docker using docker run -p 7860:7860 -e LANGFLOW_SUPERUSER_PASSWORD langflowai/langflow:latest. For another visual agent-workflow platform, compare Dify; features and licensing differ.
Data as of : 156277 stars · 10314 forks.