ShengranHu/ADAS
A research framework for automatically inventing and combining agent building blocks via a meta-agent search algorithm.
Not currently ranked — collecting fresh signals.
star history
ADAS introduces automated design of agentic systems, a new research area focused on auto-generating powerful agent architectures. The core contribution is Meta Agent Search, where a meta-level agent iteratively writes code to construct novel agent designs based on prior discoveries. This approach treats agent development as a programmable search problem, enabling the system to invent new agent components and compositions without manual engineering.
Frequently asked
- What is ShengranHu/ADAS?
- A research framework for automatically inventing and combining agent building blocks via a meta-agent search algorithm.
- Is ADAS open source?
- Yes — ShengranHu/ADAS is open source, released under the Apache-2.0 license.
- What language is ADAS written in?
- ShengranHu/ADAS is primarily written in Python.
- How popular is ADAS?
- ShengranHu/ADAS has 1.6k stars on GitHub.
- Where can I find ADAS?
- ShengranHu/ADAS is on GitHub at https://github.com/ShengranHu/ADAS.