Skip to content

P.07 — AI · Agents · Platform

Agent Management Dashboard — Lumos

One platform to create, manage, govern and monitor agents, with an agent studio, a drag-and-drop workflow builder, MCP tools and secrets management.

Role
Associate Software Engineer · Lumos R&D
Year
2025 — 26
Company
CodeGen International · Lumos
Merit Award · National AI Awards 2026 · Best Agentic AI Solution
Agent Dashboard — Lumos
Illustration: a hierarchical workflow on a canvas: a webhook triggers a manager agent, which delegates to three agents built with LangGraph or CrewAI, then a condition routes to a webhook or another agent
CodeGen International

01 — Overview

Overview

The Agent Management Dashboard is Lumos's platform for creating, managing, governing and monitoring agents.

Its Agent Studio brings together an agent dashboard, an agent builder, a workflow builder and an execution playground. Agents connect to MCP tools, workflows run sequentially or hierarchically under a manager agent, and running agents are monitored in one place. The agents and workflows built here can be used across Lumos's other platforms and applications.

02 — How it works

From an agent to a governed workflow

Agents are configured, connected to tools, composed into workflows, tested and monitored, all in one platform.

  1. 01

    Define

    Name, goal, role, backstory, category, task description and expected output.

  2. 02

    Choose a framework

    Build each agent with LangGraph or CrewAI, and set its access permissions.

  3. 03

    Connect tools

    Attach MCP tools from connected MCP servers, or APIs converted into MCP tools.

  4. 04

    Compose

    Drag and drop sequential or hierarchical workflows, with a manager agent, webhooks and conditional logic.

  5. 05

    Test

    Save workflows and test them in the execution playground.

  6. 06

    Monitor

    Session records, live executions, token usage and error counts.

03 — Features

Build, connect, govern.

  • 01Agent Studio

    An agent dashboard, agent builder, workflow builder and execution playground.

  • 02Agent builder

    Agents are configured by name, goal, role, backstory, category, task description, expected output and connected MCP tools, built with LangGraph or CrewAI, with access permissions at the agent level.

  • 03MCP tools and API integration

    External MCP servers connect through a dedicated UI, and their tools can then be used by agents. Any API can also be converted into an MCP tool.

  • 04Workflow builder

    A canvas to drag and drop components into sequential and hierarchical workflows, with a configurable manager agent, webhooks, conditional logic and other components. Workflows are saved and tested in the execution playground.

  • 05Agent monitoring and governance

    Session records, live execution views, token usage and error counts, to monitor, govern and improve running agents.

  • 06Access permissions and secrets

    Access permissions per agent, and OpenBao for managing MCP and tool secrets and keys through the platform.

04 — Technical depth

What it's built with, layer by layer.

Frontend
Next.js
Backend
FastAPI
Database
MongoDB
Agent frameworks
LangGraph · CrewAI
Tools
MCP · API-to-MCP conversion
Secrets
OpenBao

06 — Outcome

Where it landed.

  • 2

    Agent frameworks

    LangGraph and CrewAI

  • 4

    Studio tools

    dashboard, builder, workflow builder and playground

  • 2

    Workflow types

    sequential and hierarchical

The agents and workflows created here can be used across Lumos's other platforms and applications, monitored, permissioned and with their secrets managed in one place.