# ShengranHu/ADAS

[ICLR 2025] Automated Design of Agentic Systems

Repository: https://github.com/ShengranHu/ADAS
Canonical: https://ross.abutalabs.com/products/adas
Homepage: https://www.shengranhu.com/ADAS/
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2025-01-28T06:37:51+00:00

## Health v2 (maintenance only)
Score: 24/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 3, release rhythm 35, longevity 54
- inputs: {"age_days": 763, "days_push": 582, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1631, forks 237 (observed 2026-08-28T04:05:14.066545+00:00)

## What it is
ADAS (Automated Design of Agentic Systems) is the official research code for the ICLR 2025 paper introducing Meta Agent Search, where a meta agent iteratively programs novel agent designs in Python code. It provides self-contained experiment folders for benchmark domains like ARC, DROP, MGSM, and MMLU, with an OpenAI API key as the main dependency.

## Use cases
- automatically discover new agent designs with LLMs
- run Meta Agent Search experiments on benchmark domains
- adapt agent search to a custom evaluation domain
- reproduce ICLR 2025 ADAS paper results
- research automated agentic system design
- evolve and archive LLM agent programs

## When to choose
- you are doing research on automated agent design or agent architecture search
- you want to reproduce or extend the Meta Agent Search algorithm
- you need a code-based framework where agents invent new agents programmatically

## When to avoid
- you need a production-ready agent framework for deploying chatbots or pipelines
- you want a no-code or low-code agent builder
- you lack an OpenAI API key or budget for many LLM calls
- you need a stable, well-documented library rather than research code

## Facets
- artifact type: library
- maturity: active
- function: agent-framework, llm-inference, machine-learning, prompt-engineering, developer-tools
- domain: artificial-intelligence, large-language-models
- platform: python, cli, cross-platform
- tags: adas, meta-agent-search, automated-agent-design, iclr-2025, research-code, llm-agents, code-generation, ai-agents, research, automation

## Member repositories
- ShengranHu/ADAS (main) score 24

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:14.066545+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:47:49.499367+00:00, confidence not recorded.
  - readme: https://github.com/ShengranHu/ADAS (fetched 2026-08-28T04:05:14.066545+00:00, sha d0e16a9501c2)
  - homepage: https://www.shengranhu.com/ADAS/ (fetched 2026-08-29T11:20:31.575001+00:00, sha 3b814d6c507b)
- Data as of 2026-08-30T08:39:29.467469+00:00.