AgentLaboratory
Basic Information
Agent Laboratory is an end-to-end autonomous research workflow that uses specialized large language model agents to assist human researchers in implementing research ideas. The repository provides a configurable system where multiple agents collaborate across a structured three-phase research pipeline: literature review, experimentation, and report writing. It is intended to automate repetitive and time-intensive tasks such as searching and synthesizing papers, generating experiment plans and code, running experiments, and producing LaTeX reports while allowing human oversight and custom notes. The codebase includes a Python entry script and YAML experiment configs, supports checkpoint saving and loading, and can be run with different LLM backends. It also includes support for AgentRxiv, a framework for agents to upload, retrieve, and build on one another's research, enabling cumulative agent-driven progress.
Links
App Details
-
$$$$ -
$$$$