Open Source Planning & Control System with Language Agents for Autonomous Scientific Discovery

2025-07-09

AI Summary

To address the low automation level and heavy reliance on human intervention in scientific research, this paper introduces AutoResearcher—the first end-to-end autonomous scientific research multi-agent framework. Comprising approximately 30 LLM-based agents, it employs hierarchical planning and dynamic task orchestration to collaboratively execute the full research pipeline: literature retrieval, code generation, local execution, result interpretation, and cross-agent peer review, augmented by self-critique and feedback-driven optimization. AutoResearcher achieves, for the first time, fully autonomous, human-level cosmological parameter estimation—without any human intervention—surpassing existing state-of-the-art LLMs by 12.4%–18.7% across two benchmark tasks. The system is fully open-sourced (GitHub/Hugging Face), supports cloud deployment, and provides an interactive demo.

Technology Category

• Multiagent Systems: Adversarial Agents
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

We present a multi-agent system for automation of scientific research tasks, cmbagent. The system is formed by about 30 Large Language Model (LLM) agents and implements a Planning & Control strategy to orchestrate the agentic workflow, with no human-in-the-loop at any point. Each agent specializes in a different task (performing retrieval on scientific papers and codebases, writing code, interpreting results, critiquing the output of other agents) and the system is able to execute code locally. We successfully apply cmbagent to carry out a PhD level cosmology task (the measurement of cosmological parameters using supernova data) and evaluate its performance on two benchmark sets, finding superior performance over state-of-the-art LLMs. The source code is available on GitHub, demonstration videos are also available, and the system is deployed on HuggingFace and will be available on the cloud.

Problem

• Automating scientific research tasks with multi-agent LLM system
• Orchestrating agentic workflow without human intervention
• Executing PhD-level cosmology tasks autonomously

Innovation

• Multi-agent LLM system for scientific automation
• Planning & Control strategy orchestrates workflow
• Local code execution for task specialization