bymilon/ai-agents-workflows-command-central-frontend

This repo is an open source web user interface, written in React and TypeScript, for what would eventually be a control panel to run and watch AI agents. The README is upfront that what is in the box today is only the front end. The screens look like a working operations console, but the data shown in them is faked, the buttons do not actually start or stop any agents, and the parts of the documentation that talk about backend design are described as plans rather than working code. The app is organized as a single-page dashboard with a sidebar and several views. There is a workflow canvas where you can drag and connect nodes to describe an agent workflow, an overview screen with charts of execution volume and recent runs, a registry that lists agents, a memory view for vector and memory collections, and an observability page for traces and telemetry. The look is dark, dense, and aimed at operators rather than casual users. Under the hood it uses React 19, Vite 6 as the build tool, Tailwind CSS 4 for styling with theme tokens, Zustand for state, the @xyflow/react library (React Flow) for the workflow canvas, Recharts for charts, and Bun as the package manager and dev runner. To try it locally, the README says to clone the repo, run bun install, then bun run dev, and open localhost on port 3000. The source is split by feature folders for dashboard, workflows, agents, memory, and observability, plus shared layout components. Separate markdown files describe the architecture, design choices, and a roadmap or TODO list. The repository has 1 star, is licensed under MIT, and asks contributors to follow a contributing guide and a security policy for any vulnerability reports.