路墨 LumoswitchChina edition
Open console
Agent Console · All-in-One AI Toolkit

Your Agent ConsoleAn all-in-one AI toolkit

Go from model and tool setup to real tasks in your workspace, all in one client. Connect a model, choose an agent, and extend its toolkit with Skills and MCP.

Download client

Download the client and start by connecting a model and opening a workspace.

Cloud and local models Agents, Skills, and MCP Local workspaces
Client preview
Start a task in a workspace

Connect a model, choose an agent, and work in a local project.

Model connectionAgent ConsoleLocal workspace
A workspace at a glance

Models, agents, and workspaces in one place.

Model connectionsAgent workspaceSkills and MCPWorkspaces
Review settings before applying Task artifacts stay in the workspace
Model connectionsConnect cloud providers or configure a local model service.
DeepSeekQwenDoubaoKimi
Console capabilities

Models, agents, and tools in one workspace

Connect a model, choose an agent, and set up a workspace and tools for the task. Available tools vary by agent.

  1. 01

    Model connections and local routing

    Manage model connections in the client and choose the right model for each agent.

    • Cloud models and local services
    • Configure model routing as needed
    • Manage your own keys locally
  2. 02

    Popular agents, configured together

    Find, configure, and start popular agents in one client to get going on the task at hand.

    • Manage agent settings together
    • Install and enable as needed
    • Switch agents for different tasks
  3. 03

    Give agents the tools they need

    Set up Skills and MCP services so agents can take on more tasks in your workflow.

    • Manage Skills
    • Connect MCP services
    • Combine tools by task
  4. 04

    Real work in your workspace

    Open a workspace and ask an agent to write, organize research, analyze code, or move a project forward.

    • Work with local files
    • Keep task context in one place
    • For work and creative projects
Capability details

Put an agent to work on everyday tasks

Describe a goal in your workspace and let an agent help organize information, write content, or move a project forward. Adapt these examples to get started.

Project task example

Explore a project

Read the project materials and summarize its goals, progress, and suggested next steps.

Document task example

Organize documents

Turn the meeting notes in this workspace into decisions, action items, and owners.

Writing task example

Create content

Use these materials to draft a friendly event introduction that highlights how to take part.

A little company beyond work

Lighter experiences in the same desktop client

Practical tools and lighter experiences share one desktop client. The companion and pet are separate from project-agent tasks.

QQ playbook

Big Fish · QQ companion

Connect a QQ bot and model, pair a private chat, and start a conversation in QQ.

Explore
Desktop experience

Desktop pet

Keep Lumo by your side while you work, with options to adjust its look and interactions.

Getting started
Three steps to start

Connect a model, then do one small task

  1. 01
    Connect a model

    Choose a platform model or configure an accessible cloud or local service.

  2. 02
    Choose an agent and workspace

    Choose an agent for the task and open a project or set of materials.

  3. 03
    Start a task

    Describe what you want to do and review the agent's result.

Put an agent to work on everyday tasks

Read the project materials and summarize its goals, progress, and suggested next steps.

FAQ

Questions about models, agents, and QQ

Answers about local routing, agent compatibility, QQ, and everyday use.

What is 路墨 Lumoswitch?

路墨 is the Chinese brand name for Lumoswitch. It is an AI toolkit and Agent console for connecting models, configuring agents, managing Skills and MCP, and working with code, documents, and everyday tasks in a local workspace.

Related guide: What is 路墨 Lumoswitch?
Where should I start if I am new to setup?

Download the client, connect a model, and choose an agent. Open a workspace to organize documents, summarize information, or explore a project. Add Skills and MCP services as you need them.

Related guide: Where should I start if I am new to setup?
Does local routing mean offline inference?

Local routing runs on your computer and forwards requests to the model service you configure. With a local service such as Ollama, requests can stay on your machine. When configured for a cloud model, inference still runs through that cloud service and requires network access and credentials.

Related guide: Does local routing mean offline inference?
Do Skills and MCP work with every agent?

The client provides Skills and MCP settings for supported agents. Available options vary by agent; choose one in the client to see its tools.

Related guide: Do Skills and MCP work with every agent?
How do I chat with Big Fish on QQ?

Configure your QQ bot and model in the client and follow the private-chat pairing guide. Keep the desktop app and computer running to chat with Big Fish. You can also authorize individual group chats.

Related guide: How do I chat with Big Fish on QQ?
Lumoswitch China edition

Start with one console for models and agents

Download the client, connect a model, and open a workspace. You can also browse model resources first.

Download clientBrowse model resourcesOpen console