Download
EvoAgentX is an open source framework for building, evaluating, and continuously improving LLM-based agents and multi-agent workflows. It moves beyond static pipelines by introducing a self-evolving system where agents are automatically generated, tested, and optimised through iterative feedback. Developers can define goals in natural language, while the framework handles workflow creation, execution, and refinement. Its modular architecture supports layered components for agents, workflows, evaluation, and evolution, enabling flexible experimentation and scaling. EvoAgentX integrates optimisation algorithms to refine prompts, tool usage, and workflow structures over time. This allows agents to adapt dynamically instead of relying on fixed logic. It is designed for researchers and developers who want to automate complex agent systems and improve performance through continuous learning cycles, reducing manual orchestration and enabling more efficient development.
Features
- Self-evolving agent ecosystem with iterative feedback loops
- Automated workflow generation from natural language goals
- Modular architecture with layered components for flexibility
- Built-in evaluation and optimisation of agent performance
- Integration of algorithms to refine prompts and workflows
- Supports multi-agent systems and complex task orchestration
Project Samples
