Build AI Agents Visually.
Open-source studio for LangChain, LangGraph, and DeepAgents workflows. Design visually, inspect every run, export real code.
Making agentic AI understandable.
Build, Test, Automate, Ship
The full agent lifecycle in one place.
Build
- Visual Workflow Canvas – Drag, drop, configure, and version agent graphs
- 3D Spatial Builder – Inspect workflows and execution paths in a spatial view
- DeepAgent Builder – Configure tools, middleware, files, and subagents
- Repository Knowledge – Browse code and ingest files for grounded agents
- Multi-Runtime Support – LangGraph, DeepAgents, Google ADK, and managed-agent paths
- Local & Provider Models – OpenAI, Anthropic, Gemini, Ollama, LM Studio, and more
Test
- Interactive Chat – Test agents with live streaming
- Real-Time Monitoring – Watch tools, subagents, tokens, cost, and artifacts
- Human-in-the-Loop – Approval checkpoints with LangGraph interrupt()
- Run History – Trace what happened and replay important execution context
Automate
- Cron Scheduling – Automate recurring workflows with timezone support
- Webhook Triggers – Fire workflows from external services with HMAC verification
- File Watch Triggers – Trigger on file changes with glob patterns and debounce
- Presentation Export – Generate Google Slides, PDF, or Reveal.js from results
- Background Tasks – Track long-running jobs without losing execution state
Ship
- Export to Code – Production-ready Python packages
- Streamlit UI – Auto-generated web interface
- JSON Sharing – Share workflows with one click
- Local-First – Keep keys, context, and project data under your control
- Artifact Gallery – View and bulk download generated images and files
- Versioned Workflows – Compare changes, recover prior versions, and keep iteration visible
From idea to working agent is too many steps.
Graph state, runtime choice, prompts, tools, memory, approvals, subagents. By the time everything is wired together, the system is hard to see.
LangConfig keeps the graph, runtime, tools, traces, and artifacts in one visual workspace. Build on the canvas, debug live execution, and export when the workflow is ready.
Watch Agents Think
Tool selection, subagent activity, approvals, artifacts, and outputs stream live. Debug by watching the system move.
- LangGraph event streaming
- Subagent and tool inspection
- Artifacts, tokens, and run history
Canvas to Production
Export as a Python package with Streamlit UI, share workflow JSON, or turn runs into repeatable automations.
- One-click Python export
- Streamlit UI included
- JSON configuration sharing
- Schedule, webhook, and file-watch triggers
Your Data Stays Yours
Run locally with your own API keys, Ollama, LM Studio, or provider models. Keep sensitive workflows close to your machine.
Built on LangChain & LangGraph
Not a black box—LangConfig stays close to LangChain, LangGraph, and the code you can export, inspect, and change.
Works with your models
About
LangConfig is built by Cade Russell at Ghost Peony LLC. The goal is straightforward: make multi-agent AI systems accessible to everyone, whether you're a developer prototyping your next idea or someone exploring agentic AI for the first time.
LangConfig exists to lower the barrier without hiding the machinery. Build visually, learn by seeing the run, and ship when you're ready.
MIT Licensed · Actively Maintained · Open Source
Start building in 10 minutes.
Clone the repo. Build your first workflow. Ship it.
Get Started on GitHubStay in the loop
Get notified when we ship new features, integrations, and workflows.