MoreManus
An open-source multi-agent system built natively in Python, replicating the functionality of Manus. It supports automatic task decomposition, sandbox-isolated execution, browser automation, and dynamic model configuration.
Timeline (last 90 days)
MoreManus is a multi-Agent product replicating Manus; its interface features a task list, a conversation window, and a file upload function, and the underlying scheduling Agent automatically breaks tasks into multiple subtasks and assigns them to different Agents for execution.
MoreManus supports dynamic switching between any model compatible with the OpenAI API (such as GPT, DeepSeek, Gemini, and Claude), and allows dynamic configuration of model parameters, connection to the A2A Network, and dynamic addition of MCP servers on the frontend.
MoreManus's frontend is implemented with Next.js and has a built-in VNC client; the backend control center is based on FastAPI and uses the OpenAI SDK combined with the DeepSeek model to build the core brain; inside the sandbox, a FastAPI service runs with Chrome installed for remote control via CDP, and Supervisor manages the processes.
Course learning outcomes include a complete project with at least 7,000 lines of frontend, backend, and sandbox code, more than 50 Agent business application cases, and demonstrations of over 30 MCP tools.
All Facts (4)
MoreManus is a multi-Agent product replicating Manus; its interface features a task list, a conversation window, and a file upload function, and the underlying scheduling Agent automatically breaks tasks into multiple subtasks and assigns them to different Agents for execution.
50%UnverifiedMoreManus supports dynamic switching between any model compatible with the OpenAI API (such as GPT, DeepSeek, Gemini, and Claude), and allows dynamic configuration of model parameters, connection to the A2A Network, and dynamic addition of MCP servers on the frontend.
50%UnverifiedMoreManus's frontend is implemented with Next.js and has a built-in VNC client; the backend control center is based on FastAPI and uses the OpenAI SDK combined with the DeepSeek model to build the core brain; inside the sandbox, a FastAPI service runs with Chrome installed for remote control via CDP, and Supervisor manages the processes.
50%UnverifiedCourse learning outcomes include a complete project with at least 7,000 lines of frontend, backend, and sandbox code, more than 50 Agent business application cases, and demonstrations of over 30 MCP tools.
50%