Skill-MAS
What it is
Skill-MAS is a framework that evolves a single meta-skill file (SKILL.md) to automatically design and orchestrate Multi-Agent Systems (MAS). It uses a three-stage process—Task Decomposition, Agent Engineering, and Workflow Orchestration—combined with contrastive reflection to iteratively refine the skill based on benchmark performance.
The problem it solves
The complexity and manual effort required to design, prompt, and orchestrate multiple agents and their workflows for complex tasks.
What you could use it for
- Automatically generating and executing multi-agent workflows for complex tasks.
- Iteratively optimizing agent skills across multiple benchmarks (e.g., VitaBench, HLEMath).
- Generating executable Python code for multi-agent systems from a single high-level skill description.
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.