EvoAgentX: Self-Evolving Multi-Agent Framework Automatically Optimizes Workflows Through Iterative Feedback Loops — 2.5K Stars

GitHub·medium signalXBlueskyLinkedInCopy link

EvoAgentX is an open-source Python framework where agents are not statically defined: from a single prompt it builds a multi-agent workflow, then continuously evolves it using retrieval augmentation, mutation, and guided search against your dataset and goals. The framework supports Human-in-the-Loop checkpoints and both short-term and long-term memory modules, enabling agents to learn across sessions rather than resetting. With 2.5K GitHub stars and an active survey repo (Awesome-Self-Evolving-Agents), it formalizes self-improving workflows as a primary design primitive rather than an optimization afterthought.