Awesome-Agentic-Labs

KO | EN

gitlite — search

#javascript #hacktoberfest #react #ai #machine-learning #rust #linux #docker #windows #cli #awesome-list #pytorch #security #mcp

Awesome-Agentic-Labs

(★ 169)

No description available.

File Explorer

Download Latest Version (.zip)

fivelevels.jpg
fivelevels2.jpg
README.md
README_CN.md

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Awesome Agentic Labs

[![Awesome](https://awesome.re/badge.svg?logo=stylelint)](https://awesome.re)
[![Contributions Welcome](https://img.shields.io/badge/Contributions-welcome-Green?logo=mercadopago&logoColor=white)](https://github.com/wu-yc/Awesome-Agentic-Labs/pulls)
[![GitHub Stars](https://img.shields.io/github/stars/wu-yc/Awesome-Agentic-Labs?style=social)](https://star-history.com/#wu-yc/Awesome-Agentic-Labs)
[![License](https://img.shields.io/badge/License-MIT-yellow.svg)](https://raw.githubusercontent.com/wu-yc/Awesome-Agentic-Labs/main/LICENSE)

<h3 align="center">
 🤖 The AI Infrastructure of Autonomous Science 🤖 ⭐ Give us a star if you like it! ⭐
</h3>

This repository manually collects works towards the era of **agentic labs**: the AI infrastructure of autonomous science.

Authors: Yingcheng Charles Wu (Stanford), Kelsey Fu (Princeton), Le Cong (Stanford), Mengdi Wang (Princeton)  
Contact: wuyc@stanford.edu

[🌟 Overview](#-overview) • [📖 News](#-news) • [📚 Content](#-content) • [🤝 Contributing](#-contributing) • [🇨🇳 中文](README_CN.md)

</div>

## 🌟 Overview
Agentic labs represent a transformative era in AI-driven science, where autonomous systems powered by large language models (LLMs), multimodal capabilities, and integrated platforms function as full research partners. These agents generate hypotheses, design and execute experiments, analyze data, and iteratively refine processes across domains like life sciences, chemistry, materials science, physics, and astronomy.

### Conventions
### Taxonomy
- **Type**
  - **Survey**: overview / meta-analysis / perspectives
  - **System/Agent**: end-to-end or modular agentic research systems
  - **Benchmark/Eval**: benchmarks, environments, measurement frameworks
  - **Platform**: publishing / collaboration / infrastructure platforms
- **Autonomy**
  - Extracted only when authors explicitly state it (e.g., “Level 2”). Otherwise `-`.

### Links
- **Paper** links are **canonical** (arXiv/bioRxiv/DOI/publisher) when verified.  
  When not verified yet, the link points to an **arXiv search** for fast lookup.
- **Code** links point to an **official GitHub repo** when known.  
  Otherwise the link is a **GitHub repository search** (so every row has something actionable).

## 📖 News
- **[2026/02/02]** Initial version released.
- **[2026/02/18]** Added (when available) **canonical paper links**, **exact posting/submission dates**, **GitHub links** (or GitHub search links), and a lightweight **taxonomy** (**Type** + **Autonomy**).
- **[2026/04/09]** Major update: **90+ new papers** added across all sections (Dec 2025 – Apr 2026), including new benchmarks, self-driving labs, embodied science, particle physics agents, and MCP-based platforms.
- **[2026/05/09]** Added **28 high-quality papers (Apr–May 2026)** in two passes, including the foundational **EvoMaster** evolving-agent framework, real-world self-driving labs in materials (A-Lab GPSS air-sensitive Li-halide spinels, MIND, ElementsClaw superconductors, declarative Experiment-as-Code Labs) and physics (Dr.Sai at BESIII, AI CFD Scientist), DeepMind's **AI Co-Mathematician** + AlphaEvolve human-interaction study, **AiScientist** long-horizon ML research engine, autonomous medical-imaging research (Camyla) and clinical hypothesis testing (VERITAS), Earth-science agents (ClimAgent, AutoSurrogate, multi-agent hydrodynamics), the formal verification of **Sphere Packing in 8D**, new benchmarks (LABBench2, PRL-Bench, AutoResearchBench, BioVeil MATRIX), an MCP-based quantum-HPC server, and a critical "Plausible but Wrong" study on agentic failure modes in astrophysics.

---

## 📚 Content

- [1. 📚 Surveys and Frameworks](#1--surveys-and-frameworks)
- [2. 🤖 Fully Autonomous Research Pipeline](#2--fully-autonomous-research-pipeline)
- [3. 🧬 Agentic Life Sciences Research](#3--agentic-life-sciences-research)
- [4. 🧪 Agentic Chemistry Research](#4--agentic-chemistry-research)
- [5. 💎 Agentic Materials Science Research](#5--agentic-materials-science-research)
- [6. 🌌 Agentic Physics and Astronomy Research](#6--agentic-physics-and-astronomy-research)
- [Workflow Stages](#-workflow-stages)
- [Domain Applications](#-domain-applications)
- [Key Citations](#-key-citations)
- [Related Awesome Lists](#-related-awesome-lists--repos)

---

## 🔥 Methods & Systems

### 1. 📚 Surveys and Frameworks
- (2025-08-18) **From AI for Science to Agentic Science: A Survey on Autonomous Scientific Discovery** [📝 Paper](https://arxiv.org/abs/2508.14111) [💻 Code](https://github.com/search?q=From%20AI%20for%20Science%20to%20Agentic%20Science%3A%20A%20Survey%20on%20Autonomous%20Scientific%20Discovery&type=repositories) *Integrates perspectives on autonomous systems with five core capabilities and four workflow stages across natural sciences.* (Type: Survey, Autonomy: -)
- (2025-08-28) **A Survey of Scientific Large Language Models: From Data Foundations to Agent Frontiers** [📝 Paper](https://arxiv.org/abs/2508.21148) [💻 Code](https://github.com/search?q=A%20Survey%20of%20Scientific%20Large%20Language%20Models%3A%20From%20Data%20Foundations%20to%20Agent%20Frontiers&type=repositories) *Highlights advances in scientific datasets and LLMs complementing agentic science.* (Type: Survey, Autonomy: -)
- (2025-03-12) **Agentic AI for Scientific Discovery: A Survey of Progress, Challenges, and Future Directions** [📝 Paper](https://arxiv.org/abs/2503.08979) [💻 Code](https://github.com/search?q=Agentic%20AI%20for%20Scientific%20Discovery%3A%20A%20Survey%20of%20Progress%2C%20Challenges%2C%20and%20Future%20Directions&type=repositories) *Overview of agentic AI categorizing systems and progress in chemistry, biology, and materials science.* (Type: Survey, Autonomy: -)
- (2025-09-12) **The (R)evolution of Scientific Workflows in the Agentic AI Era: Towards Autonomous Science** [📝 Paper](https://arxiv.org/abs/2509.09915) [💻 Code](https://github.com/search?q=The%20%28R%29evolution%20of%20Scientific%20Workflows%20in%20the%20Agentic%20AI%20Era%3A%20Towards%20Autonomous%20Science&type=repositories) *Explores AI agents accelerating discovery by coordinating distributed facilities.* (Type: Survey, Autonomy: -)
- (2025-10-29) **Agentic AI: A Comprehensive Survey of Architectures, Applications, and Future Directions** [📝 Paper](https://arxiv.org/abs/2510.25445) [💻 Code](https://github.com/search?q=Agentic%20AI%3A%20A%20Comprehensive%20Survey%20of%20Architectures%2C%20Applications%2C%20and%20Future%20Directions&type=repositories) *Dual-paradigm framework for symbolic and neural agentic systems across domains.* (Type: Survey, Autonomy: -)
- (2025-10-13) **Autonomous Agents for Scientific Discovery: Orchestrating Scientists, Language, Code, and Physics** [📝 Paper](https://arxiv.org/abs/2510.09901) [💻 Code](https://github.com/search?q=Autonomous%20Agents%20for%20Scientific%20Discovery%3A%20Orchestrating%20Scientists%2C%20Language%2C%20Code%2C%20and%20Physics&type=repositories) *Transforms discovery lifecycle with LLM-based agents in life sciences, chemistry, physics.* (Type: Survey, Autonomy: -)
- (2025-10-16) **Agentic Discovery: Closing the Loop with Cooperative Agents** [📝 Paper](https://arxiv.org/abs/2510.13081) [💻 Code](https://github.com/search?q=Agentic%20Discovery%3A%20Closing%20the%20Loop%20with%20Cooperative%20Agents&type=repositories) *Predicts federations of agents augmenting scientific processes.* (Type: Survey, Autonomy: -)
- (2025-11-18) **Enhancing Agentic Autonomous Scientific Discovery with Vision-Language Model Capabilities** [📝 Paper](https://arxiv.org/abs/2511.14631) [💻 Code](https://github.com/CMBAgents/cmbagent) *Generates papers spanning disciplines, accepted at Agents4Science 2025.* (Type: Survey, Autonomy: -)
- (2025-12-17) **Evaluating Large Language Models in Scientific Discovery** [📝 Paper](https://arxiv.org/abs/2512.15567) [💻 Code](https://github.com/search?q=Evaluating%20Large%20Language%20Models%20in%20Scientific%20Discovery&type=repositories) *Scenario-grounded benchmark across biology, chemistry, materials, and physics with domain-expert-defined research projects.* (Type: Benchmark/Eval, Autonomy: -)
- (2025-10-29) **A Survey of AI Scientists** [📝 Paper](https://arxiv.org/abs/2510.23045) [💻 Code](https://github.com/search?q=A%20Survey%20of%20AI%20Scientists&type=repositories) *Six-stage framework decomposing the scientific workflow; charts evolution from foundational modules to closed-loop systems. Updated Jan 2026.* (Type: Survey, Autonomy: -)
- (2025-03-31) **Towards Scientific Intelligence: A Survey of LLM-based Scientific Agents** [📝 Paper](https://arxiv.org/abs/2503.24047) [💻 Code](https://github.com/search?q=Towards%20Scientific%20Intelligence%3A%20A%20Survey%20of%20LLM-based%20Scientific%20Agents&type=repositories) *Comprehensive survey of LLM scientific agents covering hypothesis generation, experiment design, data analysis, and simulation. Updated Feb 2026.* (Type: Survey, Autonomy: -)
- (2026-01-06) **Why LLMs Aren't Scientists Yet: Lessons from Four Autonomous Research Attempts** [📝 Paper](https://arxiv.org/abs/2601.03315) [💻 Code](https://github.com/Lossfunk/ai-scientist-artefacts-v1) *Case study of six-agent LLM pipelines documenting six recurring failure modes including implementation drift and memory degradation.* (Type: Survey, Autonomy: -)
- (2026-01-25) **Agentic AI for Self-Driving Laboratories in Soft Matter: Taxonomy, Benchmarks, and Open Challenges** [📝 Paper](https://arxiv.org/abs/2601.17920) [💻 Code](https://github.com/search?q=Agentic%20AI%20for%20Self-Driving%20Laboratories&type=repositories) *Survey and taxonomy of agentic AI for self-driving labs; proposes Level 0–5 autonomy scale and benchmark templates.* (Type: Survey, Autonomy: L0-L5)
- (2026-01-11) **Benchmarking Autonomy in Scientific Experiments: A Hierarchical Taxonomy for Autonomous Large-Scale Facilities** [📝 Paper](https://arxiv.org/abs/2601.06978) [💻 Code](https://github.com/search?q=Benchmarking%20Autonomy%20in%20Scientific%20Experiments&type=repositories) *Proposes 6-level BASE scale (analogous to SAE J3016) for autonomous large-scale user facilities.* (Type: Survey, Autonomy: L0-L5)
- (2026-02-07) **Transforming Science with Large Language Models** [📝 Paper](https://arxiv.org/abs/2502.05151) [💻 Code](https://github.com/search?q=Transforming%20Science%20with%20Large%20Language%20Models&type=repositories) *44-page curated overview of LLM roles across the full scientific lifecycle. Revised Mar 2026.* (Type: Survey, Autonomy: -)
- (2026-02-04) **Agentic AI in Healthcare and Medicine: A Seven-Dimensional Taxonomy** [📝 Paper](https://arxiv.org/abs/2602.04813) [💻 Code](https://github.com/search?q=Agentic%20AI%20in%20Healthcare%20and%20Medicine&type=repositories) *Seven-dimensional taxonomy for evaluating LLM-based healthcare agents across EHR analysis to treatment planning.* (Type: Survey, Autonomy: -)
- (2026-01) **Artificial Intelligence agents for biological research: a survey** [📝 Paper](https://academic.oup.com/bib/article/27/1/bbag075/8499367) [💻 Code](https://github.com/search?q=AI%20agents%20for%20biological%20research&type=repositories) *Structured review of 115 studies spanning clinical diagnosis, molecular design, and omics analysis agents.* (Type: Survey, Autonomy: -)
- (2026-02) **Agentic AI and the rise of in silico team science in biomedical research** [📝 Paper](https://www.nature.com/articles/s41587-026-03035-1) [💻 Code](https://github.com/search?q=Agentic%20AI%20in%20silico%20team%20science&type=repositories) *Nature Biotechnology perspective on teams of AI agents accelerating biomedical research.* (Type: Survey, Autonomy: -)
- (2026-01-29) **Towards Agentic Intelligence for Materials Science** [📝 Paper](https://arxiv.org/abs/2602.00169) [💻 Code](https://github.com/search?q=Towards%20Agentic%20Intelligence%20for%20Materials%20Science&type=repositories) *81-page survey covering the full agentic materials discovery loop with a roadmap toward safety-aware systems.* (Type: Survey, Autonomy: -)
- (2026-01-02) **Materials Informatics: Emergence To Autonomous Discovery In The Age Of AI** [📝 Paper](https://arxiv.org/abs/2601.00742) [💻 Code](https://github.com/search?q=Materials%20Informatics%20Emergence%20Autonomous%20Discovery&type=repositories) *Traces materials informatics from classical physics through BO, RL, and LLM-driven self-driving laboratories.* (Type: Survey, Autonomy: -)
- (2026-01-13) **Agentic AI and Machine Learning for Accelerated Materials Discovery** [📝 Paper](https://arxiv.org/abs/2601.09027) [💻 Code](https://github.com/search?q=Agentic%20AI%20Machine%20Learning%20Accelerated%20Materials&type=repositories) *Review of agentic AI in polymer materials, flow chemistry, biosensors, and batteries.* (Type: Survey, Autonomy: -)
- (2026-02-14) **Artificial Intelligence in Spectroscopy: Advancing Chemistry from Prediction to Generation** [📝 Paper](https://arxiv.org/abs/2502.09897) [💻 Code](https://github.com/search?q=AI%20Spectroscopy%20Prediction%20Generation&type=repositories) *Survey covering ML/AI for NMR, MS, IR, Raman, UV-Vis from pattern recognition to foundation models.* (Type: Survey, Autonomy: -)
- (2026-02-26) **Chemical Reaction Engineering and Catalysis: AI/ML Workflows and Self-Driving Laboratories** [📝 Paper](https://arxiv.org/abs/2603.05526) [💻 Code](https://github.com/search?q=Chemical%20Reaction%20Engineering%20AI%20Self-Driving&type=repositories) *Review arguing for agentic AI as the organizing framework for catalysis workflows and autonomous SDL integration.* (Type: Survey, Autonomy: -)
- (2026-03-30) **Deep Research of Deep Research: From Transformer to Agent, From AI to AI for Science** [📝 Paper](https://arxiv.org/abs/2603.28361) [💻 Code](https://github.com/search?q=Deep%20Research%20of%20Deep%20Research&type=repositories) *Unifies industry "deep research" products with academic AI4Science into a single developmental framework.* (Type: Survey, Autonomy: -)
- (2025-03) **Scaling Laws in Scientific Discovery with AI and Robot Scientists** [📝 Paper](https://arxiv.org/abs/2503.22444) [💻 Code](https://github.com/search?q=Scaling%20Laws%20Scientific%20Discovery%20Robot%20Scientists&type=repositories) *Proposes Autonomous Generalist Scientist (AGS) concept combining agentic AI with embodied robotics. Revised 2026.* (Type: Survey, Autonomy: -)
- (2026-02-19) **Toward a Fully Autonomous, AI-Native Particle Accelerator** [📝 Paper](https://arxiv.org/abs/2602.17536) [💻 Code](https://github.com/search?q=AI-Native%20Particle%20Accelerator&type=repositories) *Nine research thrusts for building future particle accelerators as AI-native self-driving systems.* (Type: Survey, Autonomy: -)
- (2026-02-19) **Building an AI-Native Research Ecosystem for Experimental Particle Physics** [📝 Paper](https://arxiv.org/abs/2602.17582) [💻 Code](https://github.com/search?q=AI-Native%20Research%20Ecosystem%20Particle%20Physics&type=repositories) *Community whitepaper (456+ co-authors) on AI-native ecosystems for HL-LHC, DUNE, EIC, and FCC-ee.* (Type: Survey, Autonomy: -)
- (2026-03-23) **Exploring Self-Driving Labs for Optoelectronic Materials** [📝 Paper](https://arxiv.org/abs/2603.21704) [💻 Code](https://github.com/search?q=Self-Driving%20Labs%20Optoelectronic%20Materials&type=repositories) *Proposes an exploration-driven SDL paradigm for optoelectronic materials enabling mechanistic inference.* (Type: Survey, Autonomy: -)

### 2. 🤖 Fully Autonomous Research Pipeline

#### 2.1 🌱 Foundational Evolving Agent Frameworks
- (2026-04-19) **EvoMaster: A Foundational Evolving Agent Framework for Agentic Science at Scale** [📝 Paper](https://arxiv.org/abs/2604.17406) [💻 Code](https://github.com/sjtu-sai-agents/EvoMaster) *Domain-agnostic self-evolving agent framework that iteratively refines hypotheses, self-critiques, and accumulates knowledge across experiments; SOTA on four scientific benchmarks with deployable instances in ~100 LoC.* (Type: System/Agent, Autonomy: -)

#### 2.2 🤖 End-to-End Pipelines, Platforms & Benchmarks
- (2025-07-07) **SciMaster: Towards General-Purpose Scientific AI Agents, Part I. X-Master as Foundation: Can We Lead on Humanity's Last Exam?** [📝 Paper](https://arxiv.org/abs/2507.05241) [💻 Code](https://github.com/xmaster-project/X-Master) *Benchmark for scientific AI agents evaluating research capabilities.* (Type: Benchmark/Eval, Autonomy: -)
- (2025-05-22) **NovelSeek: When Agent Becomes the Scientist--Building Closed-Loop System from Hypothesis to Verification** [📝 Paper](https://arxiv.org/abs/2505.23009) [💻 Code](https://github.com/search?q=NovelSeek%3A%20When%20Agent%20Becomes%20the%20Scientist--Building%20Closed-Loop%20System%20from%20Hypothesis%20to%20Verification&type=repositories) *Closed-loop system for hypothesis generation and verification.* (Type: System/Agent, Autonomy: -)
- (2023-09-06) **Large language models for automated open-domain scientific hypotheses discovery** [📝 Paper](https://arxiv.org/abs/2309.02726) [💻 Code](https://github.com/ChicagoHAI/hypothesis-generation) *Automates hypothesis discovery from literature.* (Type: Paper, Autonomy: -)
- (2025-03-21) **Maps: A multi-agent framework based on big seven personality and socratic guidance for multimodal scientific problem solving** [📝 Paper](https://arxiv.org/abs/2503.16905) [💻 Code](https://github.com/search?q=Maps%3A%20A%20multi-agent%20framework%20based%20on%20big%20seven%20personality%20and%20socratic%20guidance%20for%20multimodal%20scientific%20problem%20solving&type=repositories) *Personality-driven framework for multimodal problems.* (Type: System/Agent, Autonomy: -)
- (2025-03-23) **Agentrxiv: Towards collaborative autonomous research** [📝 Paper](https://arxiv.org/abs/2503.18102) [💻 Code](https://github.com/search?q=Agentrxiv%3A%20Towards%20collaborative%20autonomous%20research&type=repositories) *Platform for collaborative agentic pipelines.* (Type: Platform, Autonomy: -)
- (2025-08-20) **aiXiv: A Next-Generation Open-Access Platform for AI-Generated Research** [📝 Paper](https://arxiv.org/abs/2508.15126) [💻 Code](https://github.com/aixiv-org) *Platform for AI-driven research generation.* (Type: Platform, Autonomy: -)
- (2025-01-07) **Dolphin: Closed-loop open-ended auto-research through thinking, practice, and feedback** [📝 Paper](https://arxiv.org/abs/2501.00212) [💻 Code](https://github.com/search?q=Dolphin%3A%20Closed-loop%20open-ended%20auto-research%20through%20thinking%2C%20practice%2C%20and%20feedback&type=repositories) *Self-improving agent for open-ended research.* (Type: Paper, Autonomy: -)
- (2025-02-26) **Towards an AI co-scientist** [📝 Paper](https://arxiv.org/abs/2502.18864) [💻 Code](https://github.com/search?q=Towards%20an%20AI%20co-scientist&type=repositories) *Framework for AI as collaborative partner.* (Type: System/Agent, Autonomy: -)
- (2024-08-12) **The AI scientist: Towards fully automated open-ended scientific discovery** [📝 Paper](https://arxiv.org/abs/2408.06292) [💻 Code](https://github.com/SakanaAI/AI-Scientist) *End-to-end automation of discovery.* (Type: System/Agent, Autonomy: -)
- (2024-11-11) **The virtual lab: AI agents design new SARS-CoV-2 nanobodies with experimental validation** [📝 Paper](https://www.biorxiv.org/content/10.1101/2024.11.11.623744v1) [💻 Code](https://github.com/zou-group/virtual-lab) *Designs and validates nanobodies. (Published in Nature (2025-07-29).)* (Type: System/Agent, Autonomy: -)
- (2025-04-03) **SpatialAgent: An autonomous AI agent for spatial biology** [📝 Paper](https://www.biorxiv.org/content/10.1101/2025.04.03.646459v1) [💻 Code](https://github.com/Genentech/SpatialAgent) *Analyzes spatial biology data.* (Type: System/Agent, Autonomy: -)
- (2025-05-30) **Biomni: A general-purpose biomedical AI agent** [📝 Paper](https://www.biorxiv.org/content/10.1101/2025.05.30.656746v1) [💻 Code](https://github.com/snap-stanford/Biomni) *Versatile agent for biomedical tasks.* (Type: System/Agent, Autonomy: -)
- (2025) **Automating exploratory proteomics research via language models** [📝 Paper](https://arxiv.org/search/?query=Automating%20exploratory%20proteomics%20research%20via%20language%20models&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Automating%20exploratory%20proteomics%20research%20via%20language%20models&type=repositories) *Automates proteomics analysis.* (Type: Paper, Autonomy: -)
- (2025) **Matpilot: An LLM-enabled AI materials scientist under the framework of human-machine collaboration** [📝 Paper](https://arxiv.org/search/?query=Matpilot%3A%20An%20LLM-enabled%20AI%20materials%20scientist%20under%20the%20framework%20of%20human-machine%20collaboration&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Matpilot%3A%20An%20LLM-enabled%20AI%20materials%20scientist%20under%20the%20framework%20of%20human-machine%20collaboration&type=repositories) *Collaborative agent for materials research.* (Type: System/Agent, Autonomy: -)
- (2024) **Tora: A tool-integrated reasoning agent for mathematical problem solving** [📝 Paper](https://arxiv.org/search/?query=Tora%3A%20A%20tool-integrated%20reasoning%20agent%20for%20mathematical%20problem%20solving&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Tora%3A%20A%20tool-integrated%20reasoning%20agent%20for%20mathematical%20problem%20solving&type=repositories) *Tool-augmented agent for math in science.* (Type: System/Agent, Autonomy: -)
- (2025-07-01) **STELLA: Self-Evolving LLM Agent for Biomedical Research** [📝 Paper](https://arxiv.org/abs/2507.02004) [💻 Code](https://github.com/zaixizhang/STELLA) *Adaptive agent for bioinformatics.* (Type: System/Agent, Autonomy: -)
- (2025) **Two heads are better than one: A multi-agent system has the potential to improve scientific idea generation** [📝 Paper](https://arxiv.org/search/?query=Two%20heads%20are%20better%20than%20one%3A%20A%20multi-agent%20system%20has%20the%20potential%20to%20improve%20scientific%20idea%20generation&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Two%20heads%20are%20better%20than%20one%3A%20A%20multi-agent%20system%20has%20the%20potential%20to%20improve%20scientific%20idea%20generation&type=repositories) *Multi-agent for idea enhancement.* (Type: System/Agent, Autonomy: -)
- (2025) **Dora AI scientist: Multi-agent virtual research team for scientific exploration discovery and automated report generation** [📝 Paper](https://arxiv.org/search/?query=Dora%20AI%20scientist%3A%20Multi-agent%20virtual%20research%20team%20for%20scientific%20exploration%20discovery%20and%20automated%20report%20generation&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Dora%20AI%20scientist%3A%20Multi-agent%20virtual%20research%20team%20for%20scientific%20exploration%20discovery%20and%20automated%20report%20generation&type=repositories) *Virtual team for exploration.* (Type: System/Agent, Autonomy: -)
- (2024-06-10) **DiscoveryWorld: A virtual environment for developing and evaluating automated scientific discovery agents** [📝 Paper](https://arxiv.org/abs/2406.06769) [💻 Code](https://github.com/allenai/discoveryworld) *Benchmark for discovery agents.* (Type: Benchmark/Eval, Autonomy: -)
- (2023-12-20) **Autonomous chemical research with large language models** [📝 Paper](https://doi.org/10.1038/s41586-023-06792-0) [💻 Code](https://github.com/gomesgroup/coscientist) *LLM-driven chemistry research.* (Type: Paper, Autonomy: -)
- (2024-04-11) **ResearchAgent: Iterative research idea generation over scientific literature with large language models** [📝 Paper](https://arxiv.org/abs/2404.07738) [💻 Code](https://github.com/JinheonBaek/ResearchAgent) *Iterative idea generation.* (Type: System/Agent, Autonomy: -)
- (2025) **Agent laboratory: Using LLM agents as research assistants** [📝 Paper](https://arxiv.org/search/?query=Agent%20laboratory%3A%20Using%20LLM%20agents%20as%20research%20assistants&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Agent%20laboratory%3A%20Using%20LLM%20agents%20as%20research%20assistants&type=repositories) *Agents in lab settings.* (Type: System/Agent, Autonomy: -)
- (2024) **Agent hospital: A simulacrum of hospital with evolvable medical agents** [📝 Paper](https://arxiv.org/search/?query=Agent%20hospital%3A%20A%20simulacrum%20of%20hospital%20with%20evolvable%20medical%20agents&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Agent%20hospital%3A%20A%20simulacrum%20of%20hospital%20with%20evolvable%20medical%20agents&type=repositories) *Simulation for medical evolution.* (Type: System/Agent, Autonomy: -)
- (2024) **Conversational health agents: A personalized LLM-powered agent framework** [📝 Paper](https://arxiv.org/search/?query=Conversational%20health%20agents%3A%20A%20personalized%20LLM-powered%20agent%20framework&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Conversational%20health%20agents%3A%20A%20personalized%20LLM-powered%20agent%20framework&type=repositories) *Personalized health agents.* (Type: System/Agent, Autonomy: -)
- (2025) **An automatic end-to-end chemical synthesis development platform powered by large language models** [📝 Paper](https://arxiv.org/search/?query=An%20automatic%20end-to-end%20chemical%20synthesis%20development%20platform%20powered%20by%20large%20language%20models&searchtype=all&source=header) [💻 Code](https://github.com/search?q=An%20automatic%20end-to-end%20chemical%20synthesis%20development%20platform%20powered%20by%20large%20language%20models&type=repositories) *End-to-end synthesis platform.* (Type: Platform, Autonomy: -)
- (2025) **AlphaEvolve: A coding agent for scientific and algorithmic discovery** [📝 Paper](https://arxiv.org/search/?query=AlphaEvolve%3A%20A%20coding%20agent%20for%20scientific%20and%20algorithmic%20discovery&searchtype=all&source=header) [💻 Code](https://github.com/search?q=AlphaEvolve%3A%20A%20coding%20agent%20for%20scientific%20and%20algorithmic%20discovery&type=repositories) *Discovers algorithms in science.* (Type: System/Agent, Autonomy: -)
- (2025) **Accelerated end-to-end chemical synthesis development with large language models** [📝 Paper](https://arxiv.org/search/?query=Accelerated%20end-to-end%20chemical%20synthesis%20development%20with%20large%20language%20models&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Accelerated%20end-to-end%20chemical%20synthesis%20development%20with%20large%20language%20models&type=repositories) *Accelerates synthesis.* (Type: Paper, Autonomy: -)
- (2025) **DeepAnalyze: Agentic Large Language Models for Autonomous Data Science** [📝 Paper](https://arxiv.org/search/?query=DeepAnalyze%3A%20Agentic%20Large%20Language%20Models%20for%20Autonomous%20Data%20Science&searchtype=all&source=header) [💻 Code](https://github.com/search?q=DeepAnalyze%3A%20Agentic%20Large%20Language%20Models%20for%20Autonomous%20Data%20Science&type=repositories) *Autonomous data science with LLMs.* (Type: System/Agent, Autonomy: -)
- (2025-04-10) **The AI Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Search** [📝 Paper](https://arxiv.org/abs/2504.08066) [💻 Code](https://github.com/SakanaAI/AI-Scientist-v2) *End-to-end AI-generated peer-reviewed papers.* (Type: System/Agent, Autonomy: -)
- (2025-11-04) **Kosmos: An AI Scientist for Autonomous Discovery** [📝 Paper](https://arxiv.org/abs/2511.02824) [💻 Code](https://github.com/search?q=Kosmos%3A%20An%20AI%20Scientist%20for%20Autonomous%20Discovery&type=repositories) *Automates cycles of literature search, hypothesis, and analysis.* (Type: System/Agent, Autonomy: -)
- (2025-11-10) **AgenticSciML: Collaborative Multi-Agent Systems for Emergent Discovery in Scientific Machine Learning** [📝 Paper](https://arxiv.org/abs/2511.07262) [💻 Code](https://github.com/search?q=AgenticSciML%3A%20Collaborative%20Multi-Agent%20Systems%20for%20Emergent%20Discovery%20in%20Scientific%20Machine%20Learning&type=repositories) *Multi-agent proposing and critiquing models.* (Type: System/Agent, Autonomy: -)
- (2025-05-19) **Robin: A multi-agent system for automating scientific discovery** [📝 Paper](https://arxiv.org/abs/2505.13400) [💻 Code](https://github.com/search?q=Robin%3A%20A%20multi-agent%20system%20for%20automating%20scientific%20discovery&type=repositories) *Automates key intellectual steps.* (Type: System/Agent, Autonomy: -)
- (2025-12-25) **UniLabOS: An AI-Native Operating System for Autonomous Laboratories** [📝 Paper](https://arxiv.org/abs/2512.21766) [💻 Code](https://github.com/search?q=UniLabOS&type=repositories) *Bridges digital decision-making with embodied lab experimentation via typed stateful abstractions and CRUTD protocol.* (Type: Platform, Autonomy: -)
- (2025-12-30) **SCP: Accelerating Discovery with a Global Web of Autonomous Scientific Agents** [📝 Paper](https://arxiv.org/abs/2512.24189) [💻 Code](https://github.com/search?q=Science%20Context%20Protocol&type=repositories) *Open standard (Science Context Protocol) for federated networks of autonomous agents to describe and invoke scientific resources.* (Type: Platform, Autonomy: -)
- (2025-12-25) **SAGA: Accelerating Scientific Discovery with Autonomous Goal-Evolving Agents** [📝 Paper](https://arxiv.org/abs/2512.21782) [💻 Code](https://github.com/search?q=SAGA%20Goal-Evolving%20Agents&type=repositories) *Bi-level architecture: outer LLM loop proposes objectives, inner loop optimizes solutions across antibiotics, nanobodies, DNA, and materials.* (Type: System/Agent, Autonomy: -)
- (2026-01-04) **HeurekaBench: A Benchmarking Framework for AI Co-scientist** [📝 Paper](https://arxiv.org/abs/2601.01678) [💻 Code](https://github.com/mlbio-epfl/HeurekaBench) *Framework for open-ended exploratory benchmarks; instantiated in single-cell biology with 41 validated insights. (ICLR 2026.)* (Type: Benchmark/Eval, Autonomy: -)
- (2026-01-07) **O-Researcher: An Open Ended Deep Research Model via Multi-Agent Distillation and Agentic RL** [📝 Paper](https://arxiv.org/abs/2601.03743) [💻 Code](https://github.com/search?q=O-Researcher&type=repositories) *Trains open-source models to match closed-source deep research quality via multi-agent synthetic data and agentic RL.* (Type: System/Agent, Autonomy: -)
- (2026-01-14) **DeepResearchEval: Automated Framework for Deep Research Task Construction and Agentic Evaluation** [📝 Paper](https://arxiv.org/abs/2601.09688) [💻 Code](https://github.com/Infinity-AILab/DeepResearchEval) *Persona-driven pipeline generating multi-source research tasks with agentic evaluator using adaptive quality assessment.* (Type: Benchmark/Eval, Autonomy: -)
- (2026-01-15) **DR-Arena: An Automated Evaluation Framework for Deep Research Agents** [📝 Paper](https://arxiv.org/abs/2601.10504) [💻 Code](https://github.com/search?q=DR-Arena&type=repositories) *Builds Information Trees from live web trends for evaluation; 0.94 Spearman correlation with LMSYS Search Arena.* (Type: Benchmark/Eval, Autonomy: -)
- (2026-01-18) **Rethinking the AI Scientist: Interactive Multi-Agent Workflows for Scientific Discovery** [📝 Paper](https://arxiv.org/abs/2601.12542) [💻 Code](https://github.com/search?q=Rethinking%20AI%20Scientist%20Deep%20Research&type=repositories) *"Deep Research" multi-agent system with persistent world state; 48.8% open-response accuracy on BixBench.* (Type: System/Agent, Autonomy: -)
- (2026-01-18) **A Cloud-based Multi-Agentic Workflow for Science** [📝 Paper](https://arxiv.org/abs/2601.12607) [💻 Code](https://github.com/search?q=Cloud-based%20Multi-Agentic%20Workflow%20Science&type=repositories) *Supervisor agent orchestrates six specialized agents; 97.5% task completion on synthetic tasks; demonstrated on catalyst discovery.* (Type: System/Agent, Autonomy: -)
- (2026-01-28) **MADE: Benchmark Environments for Closed-Loop Materials Discovery** [📝 Paper](https://arxiv.org/abs/2601.20996) [💻 Code](https://github.com/search?q=MADE%20Benchmark%20Materials%20Discovery&type=repositories) *Formalizes materials discovery as constrained closed-loop search; supports fully agentic LLM systems.* (Type: Benchmark/Eval, Autonomy: -)
- (2026-02-02) **S1-NexusAgent: A Self-Evolving Agent Framework for Multidisciplinary Scientific Research** [📝 Paper](https://arxiv.org/abs/2602.01550) [💻 Code](https://github.com/CASIA-LM/S1-NexusAgent) *Hierarchical Plan-and-CodeAct agent supporting MCP with thousands of cross-disciplinary scientific tools.* (Type: System/Agent, Autonomy: -)
- (2026-02-02) **FIRE-Bench: Evaluating Agents on the Rediscovery of Scientific Insights** [📝 Paper](https://arxiv.org/abs/2602.02905) [💻 Code](https://github.com/search?q=FIRE-Bench&type=repositories) *Agents must autonomously rediscover findings from published ML studies; best agents achieve under 50 F1.* (Type: Benchmark/Eval, Autonomy: -)
- (2026-02-03) **Aster: Autonomous Scientific Discovery over 20x Faster Than Existing Methods** [📝 Paper](https://arxiv.org/abs/2602.07040) [💻 Code](https://github.com/search?q=Aster%20Autonomous%20Scientific%20Discovery&type=repositories) *Iterative program-improvement agent reducing discovery iterations by 20x; applied to math, GPU kernels, biology, and neuroscience.* (Type: System/Agent, Autonomy: -)
- (2026-02-09) **InternAgent-1.5: A Unified Agentic Framework for Long-Horizon Autonomous Scientific Discovery** [📝 Paper](https://arxiv.org/abs/2602.08990) [💻 Code](https://github.com/search?q=InternAgent&type=repositories) *Three-subsystem architecture (generation, verification, evolution) enabling continuous discovery across earth, life, and physical sciences.* (Type: System/Agent, Autonomy: -)
- (2026-02-13) **Towards Autonomous Mathematics Research (Aletheia)** [📝 Paper](https://arxiv.org/abs/2602.10177) [💻 Code](https://github.com/search?q=Aletheia%20Autonomous%20Mathematics&type=repositories) *Gemini Deep Think–powered math research agent; produced AI-generated paper in arithmetic geometry and solved four open Erdős Conjecture problems.* (Type: System/Agent, Autonomy: -)
- (2026-02-24) **Grounding LLMs in Scientific Discovery via Embodied Actions (EmbodiedAct)** [📝 Paper](https://arxiv.org/abs/2602.20639) [💻 Code](https://github.com/thu-coai/EmbodiedAct) *Grounds LLMs in scientific software via tight perception-execution loop; SOTA on engineering design and simulation benchmarks.* (Type: System/Agent, Autonomy: -)
- (2026-02-24) **PANGAEA-GPT: A Hierarchical Multi-Agent System for Autonomous Discovery in Geoscientific Data Archives** [📝 Paper](https://arxiv.org/abs/2602.21351) [💻 Code](https://github.com/search?q=PANGAEA-GPT&type=repositories) *Supervisor-Worker topology with data-type-aware routing across six geoscientific domains.* (Type: System/Agent, Autonomy: -)
- (2026-03-09) **EvoScientist: Towards Multi-Agent Evolving AI Scientists for End-to-End Scientific Discovery** [📝 Paper](https://arxiv.org/abs/2603.08127) [💻 Code](https://github.com/EvoScientist/EvoScientist) *Self-evolving multi-agent with dual persistent memories; all six generated papers accepted at AI Scientist track conference.* (Type: System/Agent, Autonomy: -)
- (2026-03-16) **The Agentic Researcher: A Practical Guide to AI-Assisted Research in Mathematics and Machine Learning** [📝 Paper](https://arxiv.org/abs/2603.15914) [💻 Code](https://github.com/ZIB-IOL/The-Agentic-Researcher) *Open-source framework converting CLI coding agents into sandboxed autonomous research assistants.* (Type: System/Agent, Autonomy: L4)
- (2026-03-18) **Toward Evaluation Frameworks for Multi-Agent Scientific AI Systems** [📝 Paper](https://arxiv.org/abs/2603.26718) [💻 Code](https://github.com/search?q=Evaluation%20Frameworks%20Multi-Agent%20Scientific%20AI&type=repositories) *Proposes contamination-resistant benchmark design and multi-turn evaluation for scientific multi-agent systems.* (Type: Benchmark/Eval, Autonomy: -)
- (2026-03-20) **Embodied Science: Closing the Discovery Loop with Agentic Embodied AI** [📝 Paper](https://arxiv.org/abs/2603.19782) [💻 Code](https://github.com/search?q=Embodied%20Science%20PLAD&type=repositories) *Proposes PLAD (Perception–Language–Action–Discovery) framework for embodied closed-loop scientific discovery.* (Type: System/Agent, Autonomy: -)
- (2026-03-24) **Bilevel Autoresearch: Meta-Autoresearching Itself** [📝 Paper](https://arxiv.org/abs/2603.23420) [💻 Code](https://github.com/search?q=Bilevel%20Autoresearch&type=repositories) *Outer loop meta-optimizes the inner autoresearch loop by injecting new search mechanisms at runtime; 5x improvement.* (Type: System/Agent, Autonomy: -)
- (2026-03-26) **MAGNET: Autonomous Expert Model Generation via Decentralized Autoresearch and BitNet Training** [📝 Paper](https://arxiv.org/abs/2603.25813) [💻 Code](https://github.com/search?q=MAGNET%20Decentralized%20Autoresearch&type=repositories) *Decentralized system combining autonomous ML research pipelines with ternary training and DiLoCo distributed merging.* (Type: System/Agent, Autonomy: -)
- (2026-03) **AI-Supervisor: Autonomous AI Research Supervision via a Persistent Research World Model** [📝 Paper](https://arxiv.org/abs/2603.24402) [💻 Code](https://github.com/search?q=AI-Supervisor%20Research%20World%20Model&type=repositories) *Multi-agent orchestration maintaining a continuously evolving Knowledge Graph for end-to-end research supervision.* (Type: System/Agent, Autonomy: -)
- (2026-03-30) **Towards a Medical AI Scientist** [📝 Paper](https://arxiv.org/abs/2603.28589) [💻 Code](https://github.com/search?q=Towards%20a%20Medical%20AI%20Scientist&type=repositories) *First autonomous research framework for clinical science with clinician-engineer co-reasoning and ethical review; Med-AI Bench with 171 cases.* (Type: System/Agent, Autonomy: -)
- (2026-03) **AutoResearchClaw: Fully Autonomous & Self-Evolving Research from Idea to Paper** [📝 Paper](https://github.com/aiming-lab/AutoResearchClaw) [💻 Code](https://github.com/aiming-lab/AutoResearchClaw) *Open-source 23-stage autonomous research pipeline turning a single idea into a conference-ready LaTeX paper.* (Type: Platform, Autonomy: -)
- (2026-04-01) **CORAL: Towards Autonomous Multi-Agent Evolution for Open-Ended Discovery** [📝 Paper](https://arxiv.org/abs/2604.01658) [💻 Code](https://github.com/Human-Agent-Society/CORAL) *First framework for autonomous multi-agent co-evolution; 3–10x higher improvement rates on 10 tasks.* (Type: System/Agent, Autonomy: -)
- (2026-04-01) **CliffSearch: Structured Agentic Co-Evolution over Theory and Code** [📝 Paper](https://arxiv.org/abs/2604.01210) [💻 Code](https://cliffsearch.ai/) *LLM-agent evolutionary framework where each node is a structured theory+code artifact with reviewer gates.* (Type: System/Agent, Autonomy: -)
- (2026-04-03) **FermiLink: A Unified Agent Framework for Multidomain Autonomous Scientific Simulations** [📝 Paper](https://arxiv.org/abs/2604.03460) [💻 Code](https://github.com/TaoELi/FermiLink) *Supports ~50 scientific software packages across 9 domains; reproduced 56.1% of 132 published figures.* (Type: Platform, Autonomy: -)
- (2025) **Aleks: AI powered Multi Agent System for Autonomous Scientific Discovery via Data-Driven Approaches in Plant Science** [📝 Paper](https://arxiv.org/search/?query=Aleks%3A%20AI%20powered%20Multi%20Agent%20System%20for%20Autonomous%20Scientific%20Discovery%20via%20Data-Driven%20Approaches%20in%20Plant%20Science&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Aleks%3A%20AI%20powered%20Multi%20Agent%20System%20for%20Autonomous%20Scientific%20Discovery%20via%20Data-Driven%20Approaches%20in%20Plant%20Science&type=repositories) *Accelerates discovery in plant sciences.* (Type: System/Agent, Autonomy: -)
- (2025-10-13) **SR-Scientist: Scientific Equation Discovery With Agentic AI** [📝 Paper](https://arxiv.org/abs/2510.11661) [💻 Code](https://github.com/search?q=SR-Scientist%3A%20Scientific%20Equation%20Discovery%20With%20Agentic%20AI&type=repositories) *Autonomous AI scientist for equation discovery.* (Type: System/Agent, Autonomy: -)
- (2025-10-30) **The Denario project: Deep knowledge AI agents for scientific discovery** [📝 Paper](https://arxiv.org/abs/2510.26887) [💻 Code](https://github.com/denario-sw/denario) *Multi-agent for generating papers across disciplines.* (Type: System/Agent, Autonomy: -)
- (2026-04-30) **AutoResearchBench: Benchmarking AI Agents on Complex Scientific Literature Discovery** [📝 Paper](https://arxiv.org/abs/2604.25256) [💻 Code](https://github.com/search?q=AutoResearchBench&type=repositories) *Benchmark with Deep Research (multi-step probing for a target paper) and Wide Research (collecting papers under conditions); top LLMs reach only ~9% on either track.* (Type: Benchmark/Eval, Autonomy: -)
- (2026-05-06) **PARNESS: A Paper Harness for End-to-End Automated Scientific Research with Dynamic Workflows, Full-Text Indexing, and Cross-Run Knowledge Accumulation** [📝 Paper](https://arxiv.org/abs/2605.05258) [💻 Code](https://github.com/gtrhythm/PARNESS) *Open-source harness combining declarative pipelines, full-PDF indexing, and cross-run knowledge graphs to overcome rigidity in autonomous research systems.* (Type: Platform, Autonomy: -)
- (2026-05-07) **AI Co-Mathematician: Accelerating Mathematicians with Agentic AI** [📝 Paper](https://arxiv.org/abs/2605.06651) [💻 Code](https://github.com/search?q=AI%20Co-Mathematician&type=repositories) *Google DeepMind interactive workbench supporting the full mathematical research lifecycle from ideation to theorem proving with an asynchronous stateful workspace; reaches 48% on FrontierMath Tier 4.* (Type: System/Agent, Autonomy: -)
- (2026-04-12) **Camyla: Scaling Autonomous Research in Medical Image Segmentation** [📝 Paper](https://arxiv.org/abs/2604.10696) [💻 Code](https://yifangao112.github.io/camyla-page/) *Fully autonomous research system that turns raw medical-imaging datasets into literature-grounded proposals, executable experiments, and complete manuscripts without human intervention.* (Type: System/Agent, Autonomy: -)
- (2026-04-14) **Toward Autonomous Long-Horizon Engineering for ML Research (AiScientist)** [📝 Paper](https://arxiv.org/abs/2604.13018) [💻 Code](https://github.com/AweAI-Team/AiScientist) *Hierarchical orchestration with a "File-as-Bus" workspace for multi-stage ML research; +10.54 PaperBench points over best baseline and 81.82 Any-Medal % on MLE-Bench Lite.* (Type: System/Agent, Autonomy: -)
- (2026-04-25) **A Milestone in Formalization: The Sphere Packing Problem in Dimension 8** [📝 Paper](https://arxiv.org/abs/2604.23468) [💻 Code](https://github.com/search?q=Sphere%20Packing%20Lean%208&type=repositories) *Lean formalization of Viazovska's 2016 sphere-packing proof; final verification stages completed by Math, Inc.'s autoformalization model "Gauss" — landmark for AI-assisted formal mathematics.* (Type: System/Agent, Autonomy: -)
- (2026-04-28) **Plausible but Wrong: A Case Study on Agentic Failures in Astrophysical Workflows** [📝 Paper](https://arxiv.org/abs/2604.25345) [💻 Code](https://github.com/search?q=Plausible%20but%20Wrong%20CMBAgent&type=repositories) *Evaluates CMBAgent on 18 astrophysical tasks and shows the dominant risk is silent generation of syntactically correct but physically incorrect results — essential reading on agentic failure modes.* (Type: Benchmark/Eval, Autonomy: -)
- (2026-05-03) **NORA: A Harness-Engineered Autonomous Research Agent for End-to-End Spatial Data Science** [📝 Paper](https://arxiv.org/abs/2605.02092) [💻 Code](https://github.com/search?q=NORA%20Night%20Owl%20Research%20Agent&type=repositories) *Domain-specialized multi-agent system with 21 GIScience workflow skills and custom safety guards for autonomous spatial data science research.* (Type: System/Agent, Autonomy: -)
- (2026-05-04) **EngiAgent: Fully Connected Coordination of LLM Agents for Solving Open-ended Engineering Problems with Feasible Solutions** [📝 Paper](https://arxiv.org/abs/2605.02289) [💻 Code](https://github.com/AI4Engi/EngiAgent) *Multi-agent system whose fully connected coordinator routes feedback flexibly across problem analysis, modeling, verification, solving, and evaluation agents — overcoming rigid pipeline-based reflection.* (Type: System/Agent, Autonomy: -)
- (2026-05-07) **Intentmaking and Sensemaking: Human Interaction with AI-Guided Mathematical Discovery** [📝 Paper](https://arxiv.org/abs/2605.05921) [💻 Code](https://github.com/search?q=Intentmaking%20Sensemaking%20AlphaEvolve&type=repositories) *Formative study of expert mathematicians using AlphaEvolve; identifies the intent-making ↔ sense-making cycle as the core human-AI collaboration pattern in agentic mathematical discovery.* (Type: Survey, Autonomy: -)

### 3. 🧬 Agentic Life Sciences Research
- (2025-05-30) **Biomni: A general-purpose biomedical AI agent** [📝 Paper](https://www.biorxiv.org/content/10.1101/2025.05.30.656746v1) [💻 Code](https://github.com/snap-stanford/Biomni) *Decomposes queries into multi-step plans (Level 2 autonomy).* (Type: System/Agent, Autonomy: L2)
- (2025) **m-KAILIN: A Knowledge-Driven Agentic Scientific Corpus Distillation Framework for Biomedical Large Language Models Training** [📝 Paper](https://arxiv.org/search/?query=m-KAILIN%3A%20A%20Knowledge-Driven%20Agentic%20Scientific%20Corpus%20Distillation%20Framework%20for%20Biomedical%20Large%20Language%20Models%20Training&searchtype=all&source=header) [💻 Code](https://github.com/search?q=m-KAILIN%3A%20A%20Knowledge-Driven%20Agentic%20Scientific%20Corpus%20Distillation%20Framework%20for%20Biomedical%20Large%20Language%20Models%20Training&type=repositories) *Framework for LLM training in biomedicine.* (Type: System/Agent, Autonomy: -)
- (2025-07-01) **STELLA: Self-Evolving LLM Agent for Biomedical Research** [📝 Paper](https://arxiv.org/abs/2507.02004) [💻 Code](https://github.com/zaixizhang/STELLA) *Adaptive agent with template library (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **From Intention to Implementation: Automating Biomedical Research via LLMs** [📝 Paper](https://arxiv.org/search/?query=From%20Intention%20to%20Implementation%3A%20Automating%20Biomedical%20Research%20via%20LLMs&searchtype=all&source=header) [💻 Code](https://github.com/search?q=From%20Intention%20to%20Implementation%3A%20Automating%20Biomedical%20Research%20via%20LLMs&type=repositories) *Automates from intention to execution.* (Type: Paper, Autonomy: -)
- (2025) **PiFlow: Principle-aware Scientific Discovery with Multi-Agent Collaboration** [📝 Paper](https://arxiv.org/search/?query=PiFlow%3A%20Principle-aware%20Scientific%20Discovery%20with%20Multi-Agent%20Collaboration&searchtype=all&source=header) [💻 Code](https://github.com/search?q=PiFlow%3A%20Principle-aware%20Scientific%20Discovery%20with%20Multi-Agent%20Collaboration&type=repositories) *Uncertainty reduction via multi-agent (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2024-04-03) **Empowering Biomedical Discovery with AI Agents** [📝 Paper](https://arxiv.org/abs/2404.02831) [💻 Code](https://github.com/search?q=Empowering%20Biomedical%20Discovery%20with%20AI%20Agents&type=repositories) *Perspective on agents in biomedical discovery.* (Type: System/Agent, Autonomy: -)
- (2024-05-25) **GeneAgent: Self-verification Language Agent for Gene Set Knowledge Discovery using Domain Databases** [📝 Paper](https://arxiv.org/abs/2405.16205) [💻 Code](https://github.com/ncbi-nlp/GeneAgent) *Self-verifying agent (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2024) **BioInformatics Agent (BIA): Unleashing the Power of Large Language Models to Reshape Bioinformatics Workflow** [📝 Paper](https://arxiv.org/search/?query=BioInformatics%20Agent%20%28BIA%29%3A%20Unleashing%20the%20Power%20of%20Large%20Language%20Models%20to%20Reshape%20Bioinformatics%20Workflow&searchtype=all&source=header) [💻 Code](https://github.com/search?q=BioInformatics%20Agent%20%28BIA%29%3A%20Unleashing%20the%20Power%20of%20Large%20Language%20Models%20to%20Reshape%20Bioinformatics%20Workflow&type=repositories) *Reshapes workflows (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2024-07-13) **CellAgent: An LLM-driven Multi-Agent Framework for Automated Single-cell Data Analysis** [📝 Paper](https://arxiv.org/abs/2407.09811) [💻 Code](https://github.com/lsq2wal/CellAgent) *Automates scRNA-seq (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **Toward a Team of AI-Made Scientists for Scientific Discovery from Gene Expression Data** [📝 Paper](https://arxiv.org/search/?query=Toward%20a%20Team%20of%20AI-Made%20Scientists%20for%20Scientific%20Discovery%20from%20Gene%20Expression%20Data&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Toward%20a%20Team%20of%20AI-Made%20Scientists%20for%20Scientific%20Discovery%20from%20Gene%20Expression%20Data&type=repositories) *Team-based discovery.* (Type: System/Agent, Autonomy: -)
- (2025) **GenoTEX: An LLM Agent Benchmark for Automated Gene Expression Data Analysis** [📝 Paper](https://arxiv.org/search/?query=GenoTEX%3A%20An%20LLM%20Agent%20Benchmark%20for%20Automated%20Gene%20Expression%20Data%20Analysis&searchtype=all&source=header) [💻 Code](https://github.com/search?q=GenoTEX%3A%20An%20LLM%20Agent%20Benchmark%20for%20Automated%20Gene%20Expression%20Data%20Analysis&type=repositories) *Benchmark for gene agents.* (Type: Benchmark/Eval, Autonomy: -)
- (2025) **GenoMAS: A Multi-Agent Framework for Scientific Discovery via Code-Driven Gene Expression Analysis** [📝 Paper](https://arxiv.org/search/?query=GenoMAS%3A%20A%20Multi-Agent%20Framework%20for%20Scientific%20Discovery%20via%20Code-Driven%20Gene%20Expression%20Analysis&searchtype=all&source=header) [💻 Code](https://github.com/search?q=GenoMAS%3A%20A%20Multi-Agent%20Framework%20for%20Scientific%20Discovery%20via%20Code-Driven%20Gene%20Expression%20Analysis&type=repositories) *Code-driven discovery.* (Type: System/Agent, Autonomy: -)
- (2024-04-25) **CRISPR-GPT: An LLM Agent for Automated Design of Gene-Editing Experiments** [📝 Paper](https://www.biorxiv.org/content/10.1101/2024.04.25.591003v1) [💻 Code](https://github.com/cong-lab/crispr-gpt-pub) *Designs experiments (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025-04-03) **SpatialAgent: An autonomous AI agent for spatial biology** [📝 Paper](https://www.biorxiv.org/content/10.1101/2025.04.03.646459v1) [💻 Code](https://github.com/Genentech/SpatialAgent) *Analyzes multi-million cell data (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **PhenoGraph: A Multi-Agent Framework for Phenotype-Driven Discovery in Spatial Transcriptomics Data Augmented with Knowledge Graphs** [📝 Paper](https://arxiv.org/search/?query=PhenoGraph%3A%20A%20Multi-Agent%20Framework%20for%20Phenotype-Driven%20Discovery%20in%20Spatial%20Transcriptomics%20Data%20Augmented%20with%20Knowledge%20Graphs&searchtype=all&source=header) [💻 Code](https://github.com/search?q=PhenoGraph%3A%20A%20Multi-Agent%20Framework%20for%20Phenotype-Driven%20Discovery%20in%20Spatial%20Transcriptomics%20Data%20Augmented%20with%20Knowledge%20Graphs&type=repositories) *Phenotype-driven discovery (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **BioAgents: Democratizing Bioinformatics Analysis with Multi-Agent Systems** [📝 Paper](https://arxiv.org/search/?query=BioAgents%3A%20Democratizing%20Bioinformatics%20Analysis%20with%20Multi-Agent%20Systems&searchtype=all&source=header) [💻 Code](https://github.com/search?q=BioAgents%3A%20Democratizing%20Bioinformatics%20Analysis%20with%20Multi-Agent%20Systems&type=repositories) *Democratizes analysis (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **BioMaster: Multi-Agent System for Automated Bioinformatics Analysis Workflow** [📝 Paper](https://arxiv.org/search/?query=BioMaster%3A%20Multi-Agent%20System%20for%20Automated%20Bioinformatics%20Analysis%20Workflow&searchtype=all&source=header) [💻 Code](https://github.com/search?q=BioMaster%3A%20Multi-Agent%20System%20for%20Automated%20Bioinformatics%20Analysis%20Workflow&type=repositories) *Automated workflows (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **TransAgent: Dynamizing Transcriptional Regulation Analysis via Multi-Omics-Aware AI Agent** [📝 Paper](https://arxiv.org/search/?query=TransAgent%3A%20Dynamizing%20Transcriptional%20Regulation%20Analysis%20via%20Multi-Omics-Aware%20AI%20Agent&searchtype=all&source=header) [💻 Code](https://github.com/search?q=TransAgent%3A%20Dynamizing%20Transcriptional%20Regulation%20Analysis%20via%20Multi-Omics-Aware%20AI%20Agent&type=repositories) *Multi-omics regulation (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **CompBioAgent: An LLM-Powered Agent for Single-Cell RNA-Seq Data Exploration** [📝 Paper](https://arxiv.org/search/?query=CompBioAgent%3A%20An%20LLM-Powered%20Agent%20for%20Single-Cell%20RNA-Seq%20Data%20Exploration&searchtype=all&source=header) [💻 Code](https://github.com/search?q=CompBioAgent%3A%20An%20LLM-Powered%20Agent%20for%20Single-Cell%20RNA-Seq%20Data%20Exploration&type=repositories) *Explores scRNA-seq (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **PerTurboAgent: A Self-Planning Agent for Boosting Sequential Perturb-seq Experiments** [📝 Paper](https://arxiv.org/search/?query=PerTurboAgent%3A%20A%20Self-Planning%20Agent%20for%20Boosting%20Sequential%20Perturb-seq%20Experiments&searchtype=all&source=header) [💻 Code](https://github.com/search?q=PerTurboAgent%3A%20A%20Self-Planning%20Agent%20for%20Boosting%20Sequential%20Perturb-seq%20Experiments&type=repositories) *Boosts experiments (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2024-11-06) **PROTEUS: Automating Exploratory Multiomics Research via Language Models** [📝 Paper](https://arxiv.org/abs/2411.03743) [💻 Code](https://github.com/search?q=PROTEUS%3A%20Automating%20Exploratory%20Multiomics%20Research%20via%20Language%20Models&type=repositories) *End-to-end multi-omics (Level 3).* (Type: Paper, Autonomy: L3)
- (2025) **CellVoyager: AI CompBio Agent Generates New Insights by Autonomously Analyzing Biological Data** [📝 Paper](https://arxiv.org/search/?query=CellVoyager%3A%20AI%20CompBio%20Agent%20Generates%20New%20Insights%20by%20Autonomously%20Analyzing%20Biological%20Data&searchtype=all&source=header) [💻 Code](https://github.com/search?q=CellVoyager%3A%20AI%20CompBio%20Agent%20Generates%20New%20Insights%20by%20Autonomously%20Analyzing%20Biological%20Data&type=repositories) *Generates insights (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **AstroAgents: A Multi-Agent AI for Hypothesis Generation from Mass Spectrometry Data** [📝 Paper](https://arxiv.org/search/?query=AstroAgents%3A%20A%20Multi-Agent%20AI%20for%20Hypothesis%20Generation%20from%20Mass%20Spectrometry%20Data&searchtype=all&source=header) [💻 Code](https://github.com/search?q=AstroAgents%3A%20A%20Multi-Agent%20AI%20for%20Hypothesis%20Generation%20from%20Mass%20Spectrometry%20Data&type=repositories) *Hypothesis from spectrometry (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **BioDiscoveryAgent: An AI Agent for Designing Genetic Perturbation Experiments** [📝 Paper](https://arxiv.org/search/?query=BioDiscoveryAgent%3A%20An%20AI%20Agent%20for%20Designing%20Genetic%20Perturbation%20Experiments&searchtype=all&source=header) [💻 Code](https://github.com/search?q=BioDiscoveryAgent%3A%20An%20AI%20Agent%20for%20Designing%20Genetic%20Perturbation%20Experiments&type=repositories) *Designs experiments (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **ProtAgents: Protein Discovery via Large Language Model Multi-Agent Collaborations Combining Physics and Machine Learning** [📝 Paper](https://arxiv.org/search/?query=ProtAgents%3A%20Protein%20Discovery%20via%20Large%20Language%20Model%20Multi-Agent%20Collaborations%20Combining%20Physics%20and%20Machine%20Learning&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ProtAgents%3A%20Protein%20Discovery%20via%20Large%20Language%20Model%20Multi-Agent%20Collaborations%20Combining%20Physics%20and%20Machine%20Learning&type=repositories) *Collaborative discovery (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **Sparks: Multi-Agent Artificial Intelligence Model Discovers Protein Design Principles** [📝 Paper](https://arxiv.org/search/?query=Sparks%3A%20Multi-Agent%20Artificial%20Intelligence%20Model%20Discovers%20Protein%20Design%20Principles&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Sparks%3A%20Multi-Agent%20Artificial%20Intelligence%20Model%20Discovers%20Protein%20Design%20Principles&type=repositories) *Discovers principles (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2024) **The Virtual Lab: AI Agents Design New SARS-CoV-2 Nanobodies with Experimental Validation** [📝 Paper](https://arxiv.org/search/?query=The%20Virtual%20Lab%3A%20AI%20Agents%20Design%20New%20SARS-CoV-2%20Nanobodies%20with%20Experimental%20Validation&searchtype=all&source=header) [💻 Code](https://github.com/search?q=The%20Virtual%20Lab%3A%20AI%20Agents%20Design%20New%20SARS-CoV-2%20Nanobodies%20with%20Experimental%20Validation&type=repositories) *Designs and validates (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **OriGene: A Self-Evolving Virtual Disease Biologist Automating Therapeutic Target Discovery** [📝 Paper](https://arxiv.org/search/?query=OriGene%3A%20A%20Self-Evolving%20Virtual%20Disease%20Biologist%20Automating%20Therapeutic%20Target%20Discovery&searchtype=all&source=header) [💻 Code](https://github.com/search?q=OriGene%3A%20A%20Self-Evolving%20Virtual%20Disease%20Biologist%20Automating%20Therapeutic%20Target%20Discovery&type=repositories) *Automates target discovery (Level 3).* (Type: Paper, Autonomy: L3)
- (2025) **Large Language Model Agent for Modular Task Execution in Drug Discovery** [📝 Paper](https://arxiv.org/search/?query=Large%20Language%20Model%20Agent%20for%20Modular%20Task%20Execution%20in%20Drug%20Discovery&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Large%20Language%20Model%20Agent%20for%20Modular%20Task%20Execution%20in%20Drug%20Discovery&type=repositories) *Modular execution (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **TxAgent: An AI Agent for Therapeutic Reasoning Across a Universe of Tools** [📝 Paper](https://arxiv.org/search/?query=TxAgent%3A%20An%20AI%20Agent%20for%20Therapeutic%20Reasoning%20Across%20a%20Universe%20of%20Tools&searchtype=all&source=header) [💻 Code](https://github.com/search?q=TxAgent%3A%20An%20AI%20Agent%20for%20Therapeutic%20Reasoning%20Across%20a%20Universe%20of%20Tools&type=repositories) *Therapeutic reasoning (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **Robin: A Multi-Agent System for Automating Scientific Discovery** [📝 Paper](https://arxiv.org/search/?query=Robin%3A%20A%20Multi-Agent%20System%20for%20Automating%20Scientific%20Discovery&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Robin%3A%20A%20Multi-Agent%20System%20for%20Automating%20Scientific%20Discovery&type=repositories) *Automates discovery (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **DrugAgent: Automating AI-Aided Drug Discovery Programming Through LLM Multi-Agent Collaboration** [📝 Paper](https://arxiv.org/search/?query=DrugAgent%3A%20Automating%20AI-Aided%20Drug%20Discovery%20Programming%20Through%20LLM%20Multi-Agent%20Collaboration&searchtype=all&source=header) [💻 Code](https://github.com/search?q=DrugAgent%3A%20Automating%20AI-Aided%20Drug%20Discovery%20Programming%20Through%20LLM%20Multi-Agent%20Collaboration&type=repositories) *Automates programming (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **LIDDIA: Language-Based Intelligent Drug Discovery Agent** [📝 Paper](https://arxiv.org/search/?query=LIDDIA%3A%20Language-Based%20Intelligent%20Drug%20Discovery%20Agent&searchtype=all&source=header) [💻 Code](https://github.com/search?q=LIDDIA%3A%20Language-Based%20Intelligent%20Drug%20Discovery%20Agent&type=repositories) *Language-based discovery (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **PharmAgents: Building a Virtual Pharma with Large Language Model Agents** [📝 Paper](https://arxiv.org/search/?query=PharmAgents%3A%20Building%20a%20Virtual%20Pharma%20with%20Large%20Language%20Model%20Agents&searchtype=all&source=header) [💻 Code](https://github.com/search?q=PharmAgents%3A%20Building%20a%20Virtual%20Pharma%20with%20Large%20Language%20Model%20Agents&type=repositories) *Virtual pharma (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **CLADD: RAG-Enhanced Collaborative LLM Agents for Drug Discovery** [📝 Paper](https://arxiv.org/search/?query=CLADD%3A%20RAG-Enhanced%20Collaborative%20LLM%20Agents%20for%20Drug%20Discovery&searchtype=all&source=header) [💻 Code](https://github.com/search?q=CLADD%3A%20RAG-Enhanced%20Collaborative%20LLM%20Agents%20for%20Drug%20Discovery&type=repositories) *Collaborative agents (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **Tippy: Accelerating Drug Discovery Through Agentic AI** [📝 Paper](https://arxiv.org/search/?query=Tippy%3A%20Accelerating%20Drug%20Discovery%20Through%20Agentic%20AI&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Tippy%3A%20Accelerating%20Drug%20Discovery%20Through%20Agentic%20AI&type=repositories) *Accelerates DMTA cycles (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **ACEGEN: Reinforcement Learning of Generative Chemical Agents for Drug Discovery** [📝 Paper](https://arxiv.org/search/?query=ACEGEN%3A%20Reinforcement%20Learning%20of%20Generative%20Chemical%20Agents%20for%20Drug%20Discovery&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ACEGEN%3A%20Reinforcement%20Learning%20of%20Generative%20Chemical%20Agents%20for%20Drug%20Discovery&type=repositories) *Generative design (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **Exploring Modularity of Agentic Systems for Drug Discovery** [📝 Paper](https://arxiv.org/search/?query=Exploring%20Modularity%20of%20Agentic%20Systems%20for%20Drug%20Discovery&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Exploring%20Modularity%20of%20Agentic%20Systems%20for%20Drug%20Discovery&type=repositories) *Analyzes modularity.* (Type: System/Agent, Autonomy: -)
- (2025) **DO Challenge: Can AI Agents Design and Implement Drug Discovery Pipelines?** [📝 Paper](https://arxiv.org/search/?query=DO%20Challenge%3A%20Can%20AI%20Agents%20Design%20and%20Implement%20Drug%20Discovery%20Pipelines%3F&searchtype=all&source=header) [💻 Code](https://github.com/search?q=DO%20Challenge%3A%20Can%20AI%20Agents%20Design%20and%20Implement%20Drug%20Discovery%20Pipelines%3F&type=repositories) *Benchmark for pipelines.* (Type: System/Agent, Autonomy: -)
- (2025-12) **SpatialBench: Can Agents Analyze Real-World Spatial Biology Data?** [📝 Paper](https://arxiv.org/abs/2512.21907) [💻 Code](https://github.com/search?q=SpatialBench&type=repositories) *146 verifiable problems across 5 spatial technologies and 7 task categories; base model accuracy only 20–38%.* (Type: Benchmark/Eval, Autonomy: -)
- (2024-12) **BioResearcher: From Intention To Implementation: Automating Biomedical Research via LLMs** [📝 Paper](https://arxiv.org/abs/2412.09429) [💻 Code](https://github.com/search?q=BioResearcher&type=repositories) *First end-to-end automated system for biomedical dry-lab research; 63.07% execution success across 8 research objectives.* (Type: System/Agent, Autonomy: -)
- (2026-01-09) **ClinAgent: A Five-Layer Architecture for Autonomous Clinical Trial Statistical Programming** [📝 Paper](https://www.medrxiv.org/content/10.64898/2026.01.09.26343542v1) [💻 Code](https://github.com/search?q=ClinAgent&type=repositories) *MCP-based skill layer for clinical programming capabilities including SAS dataset reading and CDISC-compliant code generation.* (Type: Platform, Autonomy: L2)
- (2026-01-17) **BiOmics: A Foundational Agent for Grounded and Autonomous Multi-omics Interpretation** [📝 Paper](https://www.biorxiv.org/content/10.64898/2026.01.17.699830v1) [💻 Code](https://github.com/search?q=BiOmics&type=repositories) *Dual-track architecture combining reasoning space and latent embedding space (350M daily-updated KG) for autonomous multi-omics path planning.* (Type: System/Agent, Autonomy: L3)
- (2026-01-20) **MARBLE: Multi-Agent Reasoning for Bioinformatics Learning and Evolution** [📝 Paper](https://arxiv.org/abs/2601.14349) [💻 Code](https://github.com/PRISM-DGU/MARBLE) *Execution-stable autonomous model refinement with literature-aware reference selection and structured debate among role-specialized agents.* (Type: System/Agent, Autonomy: L2)
- (2026-01-27) **Agentomics: An Agentic System that Autonomously Develops Novel SOTA Solutions for Biomedical ML Tasks** [📝 Paper](https://www.biorxiv.org/content/10.64898/2026.01.27.702049v1) [💻 Code](https://github.com/search?q=Agentomics&type=repositories) *End-to-end autonomous ML experimentation on 20 datasets across protein engineering, drug discovery, and regulatory genomics.* (Type: System/Agent, Autonomy: L3)
- (2026-01) **BioAgent Bench: An AI Agent Evaluation Suite for Bioinformatics** [📝 Paper](https://arxiv.org/abs/2601.21800) [💻 Code](https://github.com/search?q=BioAgent%20Bench&type=repositories) *Curated end-to-end bioinformatics tasks spanning bulk/single-cell RNA-seq, variant calling, metagenomics with robustness tests.* (Type: Benchmark/Eval, Autonomy: -)
- (2026-02) **BioMedAgent: Empowering AI data scientists using a multi-agent LLM framework with self-evolving capabilities** [📝 Paper](https://www.nature.com/articles/s41551-026-01634-6) [💻 Code](https://github.com/BOBQWERA/BioMedAgent) *Self-evolving multi-agent framework for autonomous tool chaining; 77% success on BioMed-AQA 327-task benchmark. (Nature Biomedical Engineering.)* (Type: System/Agent, Autonomy: L3)
- (2026-02-18) **KG-Orchestra: An Open-Source Multi-Agent Framework for Evidence-Based Biomedical Knowledge Graphs** [📝 Paper](https://www.biorxiv.org/content/10.64898/2026.02.18.706536v1) [💻 Code](https://github.com/search?q=KG-Orchestra&type=repositories) *Multi-agent framework for enriching biomedical knowledge graphs with cross-checking, self-correction, and drug repurposing.* (Type: Platform, Autonomy: L2)
- (2026-02-23) **The Virtual Biotech: A Multi-Agent AI Framework for Therapeutic Discovery** [📝 Paper](https://www.biorxiv.org/content/10.64898/2026.02.23.707551v1) [💻 Code](https://github.com/search?q=Virtual%20Biotech&type=repositories) *Team of specialist AI agents mirroring biotech org structure; analyzed 55,984 clinical trials for genomic features of drug target success.* (Type: System/Agent, Autonomy: L3)
- (2026-02-26) **PantheonOS: An Evolvable Multi-Agent Framework for Automatic Genomics Discovery** [📝 Paper](https://www.biorxiv.org/content/10.64898/2026.02.26.707870v1) [💻 Code](https://github.com/search?q=PantheonOS&type=repositories) *Privacy-preserving multi-agent framework with agentic code evolution; super-human performance on batch correction.* (Type: System/Agent, Autonomy: L3)
- (2026-02-26) **ChatSpatial: Schema-Enforced Agentic Orchestration for Reproducible Spatial Transcriptomics** [📝 Paper](https://www.biorxiv.org/content/10.64898/2026.02.26.708361v1) [💻 Code](https://github.com/search?q=ChatSpatial&type=repositories) *MCP-based platform unifying 60+ spatial transcriptomics methods across 15 analytical categories.* (Type: Platform, Autonomy: L2)
- (2026-02) **CRISPR-GPT for Agentic Automation of Gene-editing Experiments** [📝 Paper](https://www.nature.com/articles/s41551-025-01463-z) [💻 Code](https://github.com/cong-lab/crispr-gpt-pub) *Published in Nature Biomedical Engineering. LLM agent automating CRISPR experiment design end-to-end.* (Type: System/Agent, Autonomy: L2)
- (2026-03) **Mozi: Governed Autonomy for Drug Discovery LLM Agents** [📝 Paper](https://arxiv.org/abs/2603.03655) [💻 Code](https://github.com/search?q=Mozi%20Drug%20Discovery&type=repositories) *Dual-layer architecture (Control Plane + Workflow Plane) with human-in-the-loop checkpoints for drug discovery.* (Type: System/Agent, Autonomy: L2)
- (2026-03) **ELISA: An Interpretable Hybrid AI Agent for Expression-Grounded Discovery in Single-Cell Genomics** [📝 Paper](https://arxiv.org/abs/2603.11872) [💻 Code](https://github.com/omaruno/ELISA-An-AI-Agent-for-Expression-Grounded-Discovery-in-Single-Cell-Genomics) *Unifies scGPT expression embeddings with BioBERT retrieval and LLM interpretation for single-cell discovery.* (Type: System/Agent, Autonomy: L2)
- (2026-03) **Samyama: Open Biomedical Knowledge Graphs at Scale with AI Agent Access** [📝 Paper](https://arxiv.org/abs/2603.15080) [💻 Code](https://github.com/search?q=Samyama%20Biomedical%20Knowledge%20Graphs&type=repositories) *Three open-source biomedical KGs (118,686 nodes, 834,785 edges) with standardized AI agent access interfaces.* (Type: Platform, Autonomy: -)
- (2026-03) **Protein Design with Agent Rosetta** [📝 Paper](https://arxiv.org/abs/2603.15952) [💻 Code](https://github.com/search?q=Agent%20Rosetta&type=repositories) *LLM agent paired with RosettaScripts for protein design including non-canonical amino acids.* (Type: System/Agent, Autonomy: L2)
- (2026-02) **AutoBinder Agent: An MCP-Based Agent for End-to-End Protein Binder Design** [📝 Paper](https://arxiv.org/abs/2602.00019) [💻 Code](https://github.com/search?q=AutoBinder%20Agent&type=repositories) *MCP-based LLM agent orchestrating MaSIF, Rosetta, ProteinMPNN, and AlphaFold3 for automated binder design.* (Type: System/Agent, Autonomy: L3)
- (2026-03) **ProteinMCP: An Agentic AI Framework for Autonomous Protein Engineering** [📝 Paper](https://www.biorxiv.org/content/10.64898/2026.03.11.711149v1) [💻 Code](https://github.com/search?q=ProteinMCP&type=repositories) *Integrates 38 state-of-the-art protein design tools via MCP into a unified ecosystem.* (Type: Platform, Autonomy: L3)
- (2026-03) **stMCP: Spatial Transcriptomics with a Model Context Protocol Server** [📝 Paper](https://www.biorxiv.org/content/10.64898/2026.03.11.710153v1) [💻 Code](https://github.com/search?q=stMCP&type=repositories) *MCP framework for local natural language-driven spatial transcriptomics analysis preserving data privacy.* (Type: Platform, Autonomy: L2)
- (2026-03) **ClinicalAgents: Multi-Agent Orchestration for Clinical Decision Making with Dual-Memory** [📝 Paper](https://arxiv.org/abs/2603.26182) [💻 Code](https://github.com/search?q=ClinicalAgents&type=repositories) *Multi-agent system with dual-memory architecture for complex clinical reasoning tasks.* (Type: System/Agent, Autonomy: L2)
- (2026-03) **Llemy: User-driven agentic framework for analysis of large pathway diagrams** [📝 Paper](https://www.biorxiv.org/content/10.64898/2026.03.10.710813v2) [💻 Code](https://github.com/search?q=Llemy&type=repositories) *LLM-based system for exploring molecular interaction maps via natural language queries.* (Type: System/Agent, Autonomy: L2)
- (2026-03) **Latent-Y: A Lab-Validated Autonomous Agent for De Novo Drug Design** [📝 Paper](https://arxiv.org/search/?query=Latent-Y%20Autonomous%20Agent%20Drug%20Design&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Latent-Y&type=repositories) *Autonomous agent executing complete antibody design campaigns end-to-end from text prompts.* (Type: System/Agent, Autonomy: L3)
- (2026-03-30) **Claw4Science: A Dataset and Platform for the OpenClaw Scientific Agent Ecosystem** [📝 Paper](https://www.biorxiv.org/content/10.64898/2026.03.30.715118v1) [💻 Code](https://claw4science.org) *91 OpenClaw scientific agent projects and 2,230 skills across 34 scientific categories.* (Type: Platform, Autonomy: -)
- (2026-04) **MolClaw: An Autonomous Agent with Hierarchical Skills for Drug Molecule Evaluation** [📝 Paper](https://www.biorxiv.org/content/10.64898/2026.04.03.716272v1) [💻 Code](https://github.com/search?q=MolClaw&type=repositories) *Three-tier hierarchical skill architecture (70 skills, 30+ domain resources); introduces MolBench benchmark.* (Type: System/Agent, Autonomy: L3)
- (2026) **Benchmarking LLM-based agents for single-cell omics analysis** [📝 Paper](https://link.springer.com/article/10.1186/s13059-026-03998-z) [💻 Code](https://github.com/search?q=Benchmarking%20LLM%20agents%20single-cell%20omics&type=repositories) *Unified benchmarking with 50 diverse real-world single-cell omics tasks across multi-species and multi-technology. (Genome Biology.)* (Type: Benchmark/Eval, Autonomy: -)
- (2026-04-13) **Towards Autonomous Mechanistic Reasoning in Virtual Cells (VCR-Agent)** [📝 Paper](https://arxiv.org/abs/2604.11661) [💻 Code](https://github.com/search?q=VCR-Agent%20Virtual%20Cells&type=repositories) *Multi-agent framework integrating biologically grounded knowledge retrieval with verifier-based filtering to generate validated mechanistic action graphs; releases the VC-TRACES dataset.* (Type: System/Agent, Autonomy: L3)
- (2026-04-16) **CoDaS: AI Co-Data-Scientist for Biomarker Discovery via Wearable Sensors** [📝 Paper](https://arxiv.org/abs/2604.14615) [💻 Code](https://github.com/search?q=CoDaS%20Biomarker%20Wearable&type=repositories) *Multi-agent system that automates biomarker discovery on ~9,300 wearable participant observations; identifies 41 mental-health and 25 metabolic candidate digital biomarkers.* (Type: System/Agent, Autonomy: L3)
- (2026-04-16) **DeepER-Med: Advancing Deep Evidence-Based Research in Medicine Through Agentic AI** [📝 Paper](https://arxiv.org/abs/2604.15456) [💻 Code](https://github.com/search?q=DeepER-Med&type=repositories) *Agentic medical research workflow combining planning, agent collaboration, and evidence synthesis; releases DeepER-MedQA (100 expert-curated questions) and beats production platforms on clinical recommendation alignment.* (Type: System/Agent, Autonomy: L2)
- (2026-04-14) **LABBench2: An Improved Benchmark for AI Systems Performing Biology Research** [📝 Paper](https://arxiv.org/abs/2604.09554) [💻 Code](https://github.com/EdisonScientific/labbench2) *Successor to LAB-Bench with ~1,900 tasks for real-world biology research capabilities; frontier models drop 26–46 % on subtasks vs. v1, exposing a meaningful jump in difficulty for autonomous bio agents.* (Type: Benchmark/Eval, Autonomy: -)
- (2026-04-30) **BioVeil MATRIX: Uncovering and Categorizing Vulnerabilities of Agentic Biological AI Scientists** [📝 Paper](https://arxiv.org/abs/2605.00927) [💻 Code](https://bioveilmatrix.com/) *Demonstrates dual-use uplift in agentic biological tools and proposes a defensive taxonomy with 10 tactical categories and 22 techniques for red-teaming bio-agent biosecurity risks.* (Type: Benchmark/Eval, Autonomy: -)
- (2026-04-13) **VERITAS: Verifiable Epistemic Reasoning for Image-Derived Hypothesis Testing via Agentic Systems** [📝 Paper](https://arxiv.org/abs/2604.12144) [💻 Code](https://github.com/LucZot/veritas) *Multi-agent clinical-imaging hypothesis-testing system whose every conclusion traces through inspectable, executable outputs from analysis plan to segmentation masks to statistical code to final verdict.* (Type: System/Agent, Autonomy: L3)
- (2026-04-27) **Agentic AI Platforms for Autonomous Training and Rule Induction of Human–Human and Virus–Human Protein–Protein Interactions** [📝 Paper](https://arxiv.org/abs/2604.23924) [💻 Code](https://github.com/search?q=Agentic%20PPI%20Rule%20Induction&type=repositories) *Two agentic platforms autonomously train predictive PPI models and induce interpretable rules; final ensembles reach 87.3 % (human-human) and 86.5 % (human-virus) accuracy with rule-feature alignment to SHAP attributions.* (Type: System/Agent, Autonomy: L3)

### 4. 🧪 Agentic Chemistry Research
- (2024) **ChemCrow: Augmenting large-language models with chemistry tools** [📝 Paper](https://arxiv.org/search/?query=ChemCrow%3A%20Augmenting%20large-language%20models%20with%20chemistry%20tools&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ChemCrow%3A%20Augmenting%20large-language%20models%20with%20chemistry%20tools&type=repositories) *Multi-purpose agent for synthesis (Level 2).* (Type: Paper, Autonomy: L2)
- (2024) **A multi-agent-driven robotic AI chemist enabling autonomous chemical research on demand** [📝 Paper](https://arxiv.org/search/?query=A%20multi-agent-driven%20robotic%20AI%20chemist%20enabling%20autonomous%20chemical%20research%20on%20demand&searchtype=all&source=header) [💻 Code](https://github.com/search?q=A%20multi-agent-driven%20robotic%20AI%20chemist%20enabling%20autonomous%20chemical%20research%20on%20demand&type=repositories) *Hierarchical robotic chemist (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **MOOSE-Chem: Large Language Models for Rediscovering Unseen Chemistry Scientific Hypotheses** [📝 Paper](https://arxiv.org/search/?query=MOOSE-Chem%3A%20Large%20Language%20Models%20for%20Rediscovering%20Unseen%20Chemistry%20Scientific%20Hypotheses&searchtype=all&source=header) [💻 Code](https://github.com/search?q=MOOSE-Chem%3A%20Large%20Language%20Models%20for%20Rediscovering%20Unseen%20Chemistry%20Scientific%20Hypotheses&type=repositories) *Rediscovers hypotheses (Level 2).* (Type: Paper, Autonomy: L2)
- (2025) **MOOSE-Chem3: Toward Experiment-Guided Hypothesis Ranking via Simulated Experimental Feedback** [📝 Paper](https://arxiv.org/search/?query=MOOSE-Chem3%3A%20Toward%20Experiment-Guided%20Hypothesis%20Ranking%20via%20Simulated%20Experimental%20Feedback&searchtype=all&source=header) [💻 Code](https://github.com/search?q=MOOSE-Chem3%3A%20Toward%20Experiment-Guided%20Hypothesis%20Ranking%20via%20Simulated%20Experimental%20Feedback&type=repositories) *Ranks with feedback (Level 2).* (Type: Paper, Autonomy: L2)
- (2025) **An autonomous large language model agent for chemical literature data mining** [📝 Paper](https://arxiv.org/search/?query=An%20autonomous%20large%20language%20model%20agent%20for%20chemical%20literature%20data%20mining&searchtype=all&source=header) [💻 Code](https://github.com/search?q=An%20autonomous%20large%20language%20model%20agent%20for%20chemical%20literature%20data%20mining&type=repositories) *Mines data (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **Agent-based learning of materials datasets from the scientific literature** [📝 Paper](https://arxiv.org/search/?query=Agent-based%20learning%20of%20materials%20datasets%20from%20the%20scientific%20literature&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Agent-based%20learning%20of%20materials%20datasets%20from%20the%20scientific%20literature&type=repositories) *Builds datasets (Level 2).* (Type: Benchmark/Eval, Autonomy: L2)
- (2025) **ChemAgent: Enhancing LLMs for Chemistry and Materials Science through Tree-Search Based Tool Learning** [📝 Paper](https://arxiv.org/search/?query=ChemAgent%3A%20Enhancing%20LLMs%20for%20Chemistry%20and%20Materials%20Science%20through%20Tree-Search%20Based%20Tool%20Learning&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ChemAgent%3A%20Enhancing%20LLMs%20for%20Chemistry%20and%20Materials%20Science%20through%20Tree-Search%20Based%20Tool%20Learning&type=repositories) *Tool learning (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **ChemHAS: Hierarchical Agent Stacking for Enhancing Chemistry Tools** [📝 Paper](https://arxiv.org/search/?query=ChemHAS%3A%20Hierarchical%20Agent%20Stacking%20for%20Enhancing%20Chemistry%20Tools&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ChemHAS%3A%20Hierarchical%20Agent%20Stacking%20for%20Enhancing%20Chemistry%20Tools&type=repositories) *Enhances via stacking (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2024) **ChemToolAgent: The impact of tools on language agents for chemistry problem solving** [📝 Paper](https://arxiv.org/search/?query=ChemToolAgent%3A%20The%20impact%20of%20tools%20on%20language%20agents%20for%20chemistry%20problem%20solving&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ChemToolAgent%3A%20The%20impact%20of%20tools%20on%20language%20agents%20for%20chemistry%20problem%20solving&type=repositories) *Evaluates impact (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **Chemagent: Self-updating library in large language models improves chemical reasoning** [📝 Paper](https://arxiv.org/search/?query=Chemagent%3A%20Self-updating%20library%20in%20large%20language%20models%20improves%20chemical%20reasoning&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Chemagent%3A%20Self-updating%20library%20in%20large%20language%20models%20improves%20chemical%20reasoning&type=repositories) *Improves with libraries (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **LabUtopia: High-Fidelity Simulation and Hierarchical Benchmark for Scientific Embodied Agents** [📝 Paper](https://arxiv.org/search/?query=LabUtopia%3A%20High-Fidelity%20Simulation%20and%20Hierarchical%20Benchmark%20for%20Scientific%20Embodied%20Agents&searchtype=all&source=header) [💻 Code](https://github.com/search?q=LabUtopia%3A%20High-Fidelity%20Simulation%20and%20Hierarchical%20Benchmark%20for%20Scientific%20Embodied%20Agents&type=repositories) *Simulates agents (Level 2).* (Type: Benchmark/Eval, Autonomy: L2)
- (2024) **Cactus: Chemistry agent connecting tool usage to science** [📝 Paper](https://arxiv.org/search/?query=Cactus%3A%20Chemistry%20agent%20connecting%20tool%20usage%20to%20science&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Cactus%3A%20Chemistry%20agent%20connecting%20tool%20usage%20to%20science&type=repositories) *Connects tools (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2024) **Coscientist: Autonomous Reaction Optimization (Pd Cross-Coupling)** [📝 Paper](https://arxiv.org/search/?query=Coscientist%3A%20Autonomous%20Reaction%20Optimization%20%28Pd%20Cross-Coupling%29&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Coscientist%3A%20Autonomous%20Reaction%20Optimization%20%28Pd%20Cross-Coupling%29&type=repositories) *Optimizes reactions (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **LLM-RDF: End-to-End Synthesis Development** [📝 Paper](https://arxiv.org/search/?query=LLM-RDF%3A%20End-to-End%20Synthesis%20Development&searchtype=all&source=header) [💻 Code](https://github.com/search?q=LLM-RDF%3A%20End-to-End%20Synthesis%20Development&type=repositories) *End-to-end synthesis (Level 3).* (Type: Paper, Autonomy: L3)
- (2025) **Chemist-X: Reaction Condition Optimization** [📝 Paper](https://arxiv.org/search/?query=Chemist-X%3A%20Reaction%20Condition%20Optimization&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Chemist-X%3A%20Reaction%20Condition%20Optimization&type=repositories) *Optimizes conditions (Level 3).* (Type: Paper, Autonomy: L3)
- (2024) **ORGANA: Robotic Chemistry Experimentation** [📝 Paper](https://arxiv.org/search/?query=ORGANA%3A%20Robotic%20Chemistry%20Experimentation&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ORGANA%3A%20Robotic%20Chemistry%20Experimentation&type=repositories) *Executes syntheses (Level 2).* (Type: Paper, Autonomy: L2)
- (2025) **Dai et al.: Exploratory Synthesis With Mobile Robots** [📝 Paper](https://arxiv.org/search/?query=Dai%20et%20al.%3A%20Exploratory%20Synthesis%20With%20Mobile%20Robots&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Dai%20et%20al.%3A%20Exploratory%20Synthesis%20With%20Mobile%20Robots&type=repositories) *Mobile synthesis (Level 3).* (Type: Paper, Autonomy: L3)
- (2025) **Strieth-Kalthoff et al.: Closed-Loop Discovery of Laser Emitters** [📝 Paper](https://arxiv.org/search/?query=Strieth-Kalthoff%20et%20al.%3A%20Closed-Loop%20Discovery%20of%20Laser%20Emitters&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Strieth-Kalthoff%20et%20al.%3A%20Closed-Loop%20Discovery%20of%20Laser%20Emitters&type=repositories) *Closed-loop discovery (Level 3).* (Type: Paper, Autonomy: L3)
- (2025) **AutoChemSchematic AI: Generation of Industrial Process Diagrams** [📝 Paper](https://arxiv.org/search/?query=AutoChemSchematic%20AI%3A%20Generation%20of%20Industrial%20Process%20Diagrams&searchtype=all&source=header) [💻 Code](https://github.com/search?q=AutoChemSchematic%20AI%3A%20Generation%20of%20Industrial%20Process%20Diagrams&type=repositories) *Generates diagrams (Level 3).* (Type: Paper, Autonomy: L3)
- (2025) **ChatMOF: Generative Design of MOFs** [📝 Paper](https://arxiv.org/search/?query=ChatMOF%3A%20Generative%20Design%20of%20MOFs&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ChatMOF%3A%20Generative%20Design%20of%20MOFs&type=repositories) *Designs MOFs (Level 2).* (Type: Paper, Autonomy: L2)
- (2025) **MOFGen: De Novo Discovery of Synthesizable MOFs** [📝 Paper](https://arxiv.org/search/?query=MOFGen%3A%20De%20Novo%20Discovery%20of%20Synthesizable%20MOFs&searchtype=all&source=header) [💻 Code](https://github.com/search?q=MOFGen%3A%20De%20Novo%20Discovery%20of%20Synthesizable%20MOFs&type=repositories) *Discovers MOFs (Level 3).* (Type: Paper, Autonomy: L3)
- (2025) **OSDA Agent: De Novo Design of Molecules for Zeolites** [📝 Paper](https://arxiv.org/search/?query=OSDA%20Agent%3A%20De%20Novo%20Design%20of%20Molecules%20for%20Zeolites&searchtype=all&source=header) [💻 Code](https://github.com/search?q=OSDA%20Agent%3A%20De%20Novo%20Design%20of%20Molecules%20for%20Zeolites&type=repositories) *Designs for zeolites (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2024) **ChemReasoner: Heuristic Search for Catalyst Discovery** [📝 Paper](https://arxiv.org/search/?query=ChemReasoner%3A%20Heuristic%20Search%20for%20Catalyst%20Discovery&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ChemReasoner%3A%20Heuristic%20Search%20for%20Catalyst%20Discovery&type=repositories) *Searches catalysts (Level 3).* (Type: Paper, Autonomy: L3)
- (2020) **Horwood & Noutahi: Molecular Design via Reinforcement Learning** [📝 Paper](https://arxiv.org/search/?query=Horwood%20%26%20Noutahi%3A%20Molecular%20Design%20via%20Reinforcement%20Learning&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Horwood%20%26%20Noutahi%3A%20Molecular%20Design%20via%20Reinforcement%20Learning&type=repositories) *RL for design (Level 3).* (Type: Paper, Autonomy: L3)
- (2025) **El Agente Q: Autonomous Quantum Chemistry Workflows** [📝 Paper](https://arxiv.org/search/?query=El%20Agente%20Q%3A%20Autonomous%20Quantum%20Chemistry%20Workflows&searchtype=all&source=header) [💻 Code](https://github.com/search?q=El%20Agente%20Q%3A%20Autonomous%20Quantum%20Chemistry%20Workflows&type=repositories) *Automates workflows (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **Aitomia: Intelligent Assistant for Atomistic Simulations** [📝 Paper](https://arxiv.org/search/?query=Aitomia%3A%20Intelligent%20Assistant%20for%20Atomistic%20Simulations&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Aitomia%3A%20Intelligent%20Assistant%20for%20Atomistic%20Simulations&type=repositories) *Assists simulations (Level 2).* (Type: Paper, Autonomy: L2)
- (2025) **ChemGraph: Automated Computational Chemistry Workflows** [📝 Paper](https://arxiv.org/search/?query=ChemGraph%3A%20Automated%20Computational%20Chemistry%20Workflows&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ChemGraph%3A%20Automated%20Computational%20Chemistry%20Workflows&type=repositories) *Automates DFT/MD (Level 2).* (Type: Paper, Autonomy: L2)
- (2025) **xChemAgents: Explainable Quantum Chemistry Prediction** [📝 Paper](https://arxiv.org/search/?query=xChemAgents%3A%20Explainable%20Quantum%20Chemistry%20Prediction&searchtype=all&source=header) [💻 Code](https://github.com/search?q=xChemAgents%3A%20Explainable%20Quantum%20Chemistry%20Prediction&type=repositories) *Explainable predictions (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **MT-Mol: Multi Agent System with Tool-based Reasoning for Molecular Optimization** [📝 Paper](https://arxiv.org/search/?query=MT-Mol%3A%20Multi%20Agent%20System%20with%20Tool-based%20Reasoning%20for%20Molecular%20Optimization&searchtype=all&source=header) [💻 Code](https://github.com/search?q=MT-Mol%3A%20Multi%20Agent%20System%20with%20Tool-based%20Reasoning%20for%20Molecular%20Optimization&type=repositories) *Tool-based optimization.* (Type: System/Agent, Autonomy: -)
- (2025) **CSstep: Step-by-step exploration of the chemical space of drug molecules via multi-agent and multi-stage reinforcement learning** [📝 Paper](https://arxiv.org/search/?query=CSstep%3A%20Step-by-step%20exploration%20of%20the%20chemical%20space%20of%20drug%20molecules%20via%20multi-agent%20and%20multi-stage%20reinforcement%20learning&searchtype=all&source=header) [💻 Code](https://github.com/search?q=CSstep%3A%20Step-by-step%20exploration%20of%20the%20chemical%20space%20of%20drug%20molecules%20via%20multi-agent%20and%20multi-stage%20reinforcement%20learning&type=repositories) *Multi-stage RL for exploration.* (Type: System/Agent, Autonomy: -)
- (2025) **Agentic Mixture-of-Workflows for Multi-Modal Chemical Search** [📝 Paper](https://arxiv.org/search/?query=Agentic%20Mixture-of-Workflows%20for%20Multi-Modal%20Chemical%20Search&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Agentic%20Mixture-of-Workflows%20for%20Multi-Modal%20Chemical%20Search&type=repositories) *Multi-modal search.* (Type: System/Agent, Autonomy: -)
- (2025) **AI agents in chemical research: GVIM--an intelligent research assistant system** [📝 Paper](https://arxiv.org/search/?query=AI%20agents%20in%20chemical%20research%3A%20GVIM--an%20intelligent%20research%20assistant%20system&searchtype=all&source=header) [💻 Code](https://github.com/search?q=AI%20agents%20in%20chemical%20research%3A%20GVIM--an%20intelligent%20research%20assistant%20system&type=repositories) *Intelligent assistant.* (Type: System/Agent, Autonomy: -)
- (2025-12-18) **Synthelite: Chemist-aligned and Feasibility-aware Synthesis Planning with LLMs** [📝 Paper](https://arxiv.org/abs/2512.16424) [💻 Code](https://github.com/search?q=Synthelite&type=repositories) *MCTS-based explorative retrosynthesis framework; ~74% solve rate, up to 95% on constrained tasks.* (Type: System/Agent, Autonomy: L2)
- (2026-01-11) **From Text to Simulation: A Multi-Agent LLM Workflow for Automated Chemical Process Design** [📝 Paper](https://arxiv.org/abs/2601.06776) [💻 Code](https://github.com/search?q=Text%20to%20Simulation%20Chemical%20Process&type=repositories) *Four-agent workflow converting textual process descriptions into executable simulations; 89% faster design.* (Type: System/Agent, Autonomy: L3)
- (2026-01-15) **Large Language Model Agent for User-friendly Chemical Process Simulations** [📝 Paper](https://arxiv.org/abs/2601.11650) [💻 Code](https://github.com/search?q=LLM%20Chemical%20Process%20Simulations&type=repositories) *LLM integrated with AVEVA Process Simulation via MCP for natural language-driven process analysis.* (Type: System/Agent, Autonomy: L2)
- (2026-01-25) **ChemCRAFT: Agentic Reinforcement Learning for Chemical Language Models** [📝 Paper](https://arxiv.org/abs/2601.17687) [💻 Code](https://github.com/HowardLi1984/ChemCraft) *Trains small LLMs to orchestrate chemistry tool sandboxes via agentic RL + SMILES-GRPO, outperforming GPT-4.* (Type: System/Agent, Autonomy: L3)
- (2026-01-30) **QUASAR: A Universal Autonomous System for Atomistic Simulation** [📝 Paper](https://arxiv.org/abs/2602.00185) [💻 Code](https://github.com/search?q=QUASAR%20Atomistic%20Simulation&type=repositories) *Autonomously orchestrates DFT, ML potentials, MD, and Monte Carlo across three tiers of benchmark tasks including photocatalyst screening.* (Type: System/Agent, Autonomy: L3)
- (2026-02-07) **GraphAgents: Knowledge Graph-Guided Agentic AI for Cross-Domain Materials Design** [📝 Paper](https://arxiv.org/abs/2602.07491) [💻 Code](https://github.com/search?q=GraphAgents&type=repositories) *Multi-agent framework guided by large-scale KGs to find sustainable PFAS substitutes.* (Type: System/Agent, Autonomy: L2)
- (2026-02-10) **Beyond SMILES: Evaluating Agentic Systems for Drug Discovery** [📝 Paper](https://arxiv.org/abs/2602.10163) [💻 Code](https://github.com/search?q=Beyond%20SMILES%20Drug%20Discovery&type=repositories) *Evaluates six drug-discovery agent frameworks across 15 task classes; identifies five critical architectural gaps.* (Type: Benchmark/Eval, Autonomy: -)
- (2026-02-10) **AgentCAT: An LLM Agent for Extracting and Analyzing Catalytic Reaction Data** [📝 Paper](https://arxiv.org/abs/2602.18479) [💻 Code](https://github.com/search?q=AgentCAT&type=repositories) *Schema-governed closed-loop agent mining ~800 publications to extract catalytic reaction knowledge graphs.* (Type: System/Agent, Autonomy: L2)
- (2026-03-01) **Catalyst-Agent: Autonomous Heterogeneous Catalyst Screening and Optimization** [📝 Paper](https://arxiv.org/abs/2603.01311) [💻 Code](https://github.com/search?q=Catalyst-Agent&type=repositories) *MCP server-based LLM agent for ORR, NRR, and CO2RR catalyst screening via material databases and GNNs.* (Type: System/Agent, Autonomy: L3)
- (2026-03-13) **RetroReasoner: A Reasoning LLM for Strategic Retrosynthesis Prediction** [📝 Paper](https://arxiv.org/abs/2603.12666) [💻 Code](https://github.com/search?q=RetroReasoner&type=repositories) *SFT + RL-trained retrosynthesis model generating structured disconnection rationales alongside reactant predictions.* (Type: System/Agent, Autonomy: L2)
- (2026-03-31) **SimMOF: AI Agent for Automated MOF Simulations** [📝 Paper](https://arxiv.org/abs/2603.29152) [💻 Code](https://github.com/search?q=SimMOF&type=repositories) *Multi-agent framework translating natural language MOF simulation requests into dependency-aware plans.* (Type: System/Agent, Autonomy: L2)
- (2026-04-02) **PolyJarvis: LLM Agent for Autonomous Polymer MD Simulations** [📝 Paper](https://arxiv.org/abs/2604.02537) [💻 Code](https://github.com/search?q=PolyJarvis&type=repositories) *Couples LLM with RadonPy via MCP to fully automate polymer property prediction from SMILES string.* (Type: System/Agent, Autonomy: L3)
- (2025-04) **MOFGen: System of Agentic AI for the Discovery of Metal-Organic Frameworks** [📝 Paper](https://arxiv.org/abs/2504.14110) [💻 Code](https://github.com/search?q=MOFGen&type=repositories) *Multi-agent pipeline generating hundreds of thousands of novel MOFs; five experimentally synthesized for CO2 capture. (Yaghi lab, Berkeley.)* (Type: System/Agent, Autonomy: L3)
- (2026) **LFAST: Synthesis of Highly Crystalline COFs Using Large Language Models** [📝 Paper](https://pubs.acs.org/doi/10.1021/jacs.5c23233) [💻 Code](https://github.com/search?q=LFAST%20COF&type=repositories) *LLM agent mining COF literature and designing iterative 96-well experiments; 350% crystallinity boost. (JACS, Yaghi lab.)* (Type: System/Agent, Autonomy: L3)
- (2026) **MOSAIC: Collective Intelligence for AI-Assisted Chemical Synthesis** [📝 Paper](https://www.nature.com/articles/s41586-026-10131-4) [💻 Code](https://github.com/haoteli/MOSAIC) *2,498 Llama-3.1-8B specialist models generating reproducible protocols; 71% success across 35+ newly synthesized compounds. (Nature.)* (Type: Platform, Autonomy: L2)
- (2026-05-06) **Agentic Discovery of Exchange–Correlation Density Functionals** [📝 Paper](https://arxiv.org/abs/2605.05460) [💻 Code](https://github.com/search?q=Agentic%20Discovery%20Exchange%20Correlation%20Functionals&type=repositories) *LLM agent proposes structured XC-functional form changes inside a plan–execute–summarize loop; the discovered SAFS26-a improves on the gold-standard ωB97M-V baseline by ~9% on a thermochemistry benchmark.* (Type: System/Agent, Autonomy: L3)

### 5. 💎 Agentic Materials Science Research
- (2025) **AILA: AFM Nanocharacterization** [📝 Paper](https://arxiv.org/search/?query=AILA%3A%20AFM%20Nanocharacterization&searchtype=all&source=header) [💻 Code](https://github.com/search?q=AILA%3A%20AFM%20Nanocharacterization&type=repositories) *Nanocharacterization (Level 2).* (Type: Paper, Autonomy: L2)
- (2025) **Foam-Agent: Towards Automated Intelligent CFD Workflows** [📝 Paper](https://arxiv.org/search/?query=Foam-Agent%3A%20Towards%20Automated%20Intelligent%20CFD%20Workflows&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Foam-Agent%3A%20Towards%20Automated%20Intelligent%20CFD%20Workflows&type=repositories) *Automates CFD (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **ChemGraph: Computational Chemistry (DFT, MD)** [📝 Paper](https://arxiv.org/search/?query=ChemGraph%3A%20Computational%20Chemistry%20%28DFT%2C%20MD%29&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ChemGraph%3A%20Computational%20Chemistry%20%28DFT%2C%20MD%29&type=repositories) *Simulations (Level 2).* (Type: Paper, Autonomy: L2)
- (2025) **MechAgents: Large language model multi-agent collaborations can solve mechanics problems, generate new data, and integrate knowledge** [📝 Paper](https://arxiv.org/search/?query=MechAgents%3A%20Large%20language%20model%20multi-agent%20collaborations%20can%20solve%20mechanics%20problems%2C%20generate%20new%20data%2C%20and%20integrate%20knowledge&searchtype=all&source=header) [💻 Code](https://github.com/search?q=MechAgents%3A%20Large%20language%20model%20multi-agent%20collaborations%20can%20solve%20mechanics%20problems%2C%20generate%20new%20data%2C%20and%20integrate%20knowledge&type=repositories) *Solves mechanics (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2024-11-10) **MatPilot: An LLM-enabled AI materials scientist under the framework of human-machine collaboration** [📝 Paper](https://arxiv.org/abs/2411.07223) [💻 Code](https://github.com/search?q=MatPilot%3A%20An%20LLM-enabled%20AI%20materials%20scientist%20under%20the%20framework%20of%20human-machine%20collaboration&type=repositories) *Collaborative discovery (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2024) **LLMatDesign: Autonomous materials discovery with large language models** [📝 Paper](https://arxiv.org/search/?query=LLMatDesign%3A%20Autonomous%20materials%20discovery%20with%20large%20language%20models&searchtype=all&source=header) [💻 Code](https://github.com/search?q=LLMatDesign%3A%20Autonomous%20materials%20discovery%20with%20large%20language%20models&type=repositories) *LLM-driven design (Level 3).* (Type: Paper, Autonomy: L3)
- (2025) **MAPPS: Toward Greater Autonomy in Materials Discovery Agents: Unifying Planning, Physics, and Scientists** [📝 Paper](https://arxiv.org/search/?query=MAPPS%3A%20Toward%20Greater%20Autonomy%20in%20Materials%20Discovery%20Agents%3A%20Unifying%20Planning%2C%20Physics%2C%20and%20Scientists&searchtype=all&source=header) [💻 Code](https://github.com/search?q=MAPPS%3A%20Toward%20Greater%20Autonomy%20in%20Materials%20Discovery%20Agents%3A%20Unifying%20Planning%2C%20Physics%2C%20and%20Scientists&type=repositories) *Autonomous platform (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **LLaMP: Large language model made powerful for high-fidelity materials knowledge retrieval and distillation** [📝 Paper](https://arxiv.org/search/?query=LLaMP%3A%20Large%20language%20model%20made%20powerful%20for%20high-fidelity%20materials%20knowledge%20retrieval%20and%20distillation&searchtype=all&source=header) [💻 Code](https://github.com/search?q=LLaMP%3A%20Large%20language%20model%20made%20powerful%20for%20high-fidelity%20materials%20knowledge%20retrieval%20and%20distillation&type=repositories) *RAG-enhanced informatics (Level 2).* (Type: Paper, Autonomy: L2)
- (2025) **HoneyComb: A flexible LLM-based agent system for materials science** [📝 Paper](https://arxiv.org/search/?query=HoneyComb%3A%20A%20flexible%20LLM-based%20agent%20system%20for%20materials%20science&searchtype=all&source=header) [💻 Code](https://github.com/search?q=HoneyComb%3A%20A%20flexible%20LLM-based%20agent%20system%20for%20materials%20science&type=repositories) *Knowledge base (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **Multicrossmodal Automated Agent for Integrating Diverse Materials Science Data** [📝 Paper](https://arxiv.org/search/?query=Multicrossmodal%20Automated%20Agent%20for%20Integrating%20Diverse%20Materials%20Science%20Data&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Multicrossmodal%20Automated%20Agent%20for%20Integrating%20Diverse%20Materials%20Science%20Data&type=repositories) *Integrates data (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **PiFlow: Principle-aware Scientific Discovery with Multi-Agent Collaboration** [📝 Paper](https://arxiv.org/search/?query=PiFlow%3A%20Principle-aware%20Scientific%20Discovery%20with%20Multi-Agent%20Collaboration&searchtype=all&source=header) [💻 Code](https://github.com/search?q=PiFlow%3A%20Principle-aware%20Scientific%20Discovery%20with%20Multi-Agent%20Collaboration&type=repositories) *Uncertainty reduction (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **dZiner: Rational inverse design of materials with AI agents** [📝 Paper](https://arxiv.org/search/?query=dZiner%3A%20Rational%20inverse%20design%20of%20materials%20with%20AI%20agents&searchtype=all&source=header) [💻 Code](https://github.com/search?q=dZiner%3A%20Rational%20inverse%20design%20of%20materials%20with%20AI%20agents&type=repositories) *Inverse design (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **Hypothesis Generation for Materials Discovery and Design Using Goal-Driven and Constraint-Guided LLM Agents** [📝 Paper](https://arxiv.org/search/?query=Hypothesis%20Generation%20for%20Materials%20Discovery%20and%20Design%20Using%20Goal-Driven%20and%20Constraint-Guided%20LLM%20Agents&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Hypothesis%20Generation%20for%20Materials%20Discovery%20and%20Design%20Using%20Goal-Driven%20and%20Constraint-Guided%20LLM%20Agents&type=repositories) *Generates hypotheses (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **AutoMat: Enabling Automated Crystal Structure Reconstruction from Microscopy via Agentic Tool Use** [📝 Paper](https://arxiv.org/search/?query=AutoMat%3A%20Enabling%20Automated%20Crystal%20Structure%20Reconstruction%20from%20Microscopy%20via%20Agentic%20Tool%20Use&searchtype=all&source=header) [💻 Code](https://github.com/search?q=AutoMat%3A%20Enabling%20Automated%20Crystal%20Structure%20Reconstruction%20from%20Microscopy%20via%20Agentic%20Tool%20Use&type=repositories) *Reconstructs structures (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **AtomAgents: Alloy design and discovery through physics-aware multi-modal multi-agent artificial intelligence** [📝 Paper](https://arxiv.org/search/?query=AtomAgents%3A%20Alloy%20design%20and%20discovery%20through%20physics-aware%20multi-modal%20multi-agent%20artificial%20intelligence&searchtype=all&source=header) [💻 Code](https://github.com/search?q=AtomAgents%3A%20Alloy%20design%20and%20discovery%20through%20physics-aware%20multi-modal%20multi-agent%20artificial%20intelligence&type=repositories) *Designs alloys (Level 3).* (Type: System/Agent, Autonomy: L3)
- (2025) **Automating alloy design and discovery with physics-aware multimodal multiagent AI** [📝 Paper](https://arxiv.org/search/?query=Automating%20alloy%20design%20and%20discovery%20with%20physics-aware%20multimodal%20multiagent%20AI&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Automating%20alloy%20design%20and%20discovery%20with%20physics-aware%20multimodal%20multiagent%20AI&type=repositories) *Automates alloy discovery.* (Type: System/Agent, Autonomy: -)
- (2025) **Rapid and automated alloy design with graph neural network-powered LLM-driven multi-agent systems** [📝 Paper](https://arxiv.org/search/?query=Rapid%20and%20automated%20alloy%20design%20with%20graph%20neural%20network-powered%20LLM-driven%20multi-agent%20systems&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Rapid%20and%20automated%20alloy%20design%20with%20graph%20neural%20network-powered%20LLM-driven%20multi-agent%20systems&type=repositories) *Rapid design.* (Type: System/Agent, Autonomy: -)
- (2025) **metaAgent: Electromagnetic metamaterial discovery through multi-agent collaboration** [📝 Paper](https://arxiv.org/search/?query=metaAgent%3A%20Electromagnetic%20metamaterial%20discovery%20through%20multi-agent%20collaboration&searchtype=all&source=header) [💻 Code](https://github.com/search?q=metaAgent%3A%20Electromagnetic%20metamaterial%20discovery%20through%20multi-agent%20collaboration&type=repositories) *Discovers metamaterials.* (Type: System/Agent, Autonomy: -)
- (2025) **CrossMatAgent: A Multi-Agent Framework for Accelerated Metamaterial Design** [📝 Paper](https://arxiv.org/search/?query=CrossMatAgent%3A%20A%20Multi-Agent%20Framework%20for%20Accelerated%20Metamaterial%20Design&searchtype=all&source=header) [💻 Code](https://github.com/search?q=CrossMatAgent%3A%20A%20Multi-Agent%20Framework%20for%20Accelerated%20Metamaterial%20Design&type=repositories) *Accelerated design.* (Type: System/Agent, Autonomy: -)
- (2025) **An Agentic Framework for Autonomous Metamaterial Modeling and Inverse Design** [📝 Paper](https://arxiv.org/search/?query=An%20Agentic%20Framework%20for%20Autonomous%20Metamaterial%20Modeling%20and%20Inverse%20Design&searchtype=all&source=header) [💻 Code](https://github.com/search?q=An%20Agentic%20Framework%20for%20Autonomous%20Metamaterial%20Modeling%20and%20Inverse%20Design&type=repositories) *Modeling and design.* (Type: System/Agent, Autonomy: -)
- (2025) **SciAgents: Automating scientific discovery through bioinspired multi-agent intelligent graph reasoning** [📝 Paper](https://arxiv.org/search/?query=SciAgents%3A%20Automating%20scientific%20discovery%20through%20bioinspired%20multi-agent%20intelligent%20graph%20reasoning&searchtype=all&source=header) [💻 Code](https://github.com/search?q=SciAgents%3A%20Automating%20scientific%20discovery%20through%20bioinspired%20multi-agent%20intelligent%20graph%20reasoning&type=repositories) *Bioinspired discovery.* (Type: System/Agent, Autonomy: -)
- (2025) **PriM: Principle-inspired material discovery through multi-agent collaboration** [📝 Paper](https://arxiv.org/search/?query=PriM%3A%20Principle-inspired%20material%20discovery%20through%20multi-agent%20collaboration&searchtype=all&source=header) [💻 Code](https://github.com/search?q=PriM%3A%20Principle-inspired%20material%20discovery%20through%20multi-agent%20collaboration&type=repositories) *Principle-inspired discovery.* (Type: System/Agent, Autonomy: -)
- (2025) **TopoMAS: Large Language Model Driven Topological Materials Multiagent System** [📝 Paper](https://arxiv.org/search/?query=TopoMAS%3A%20Large%20Language%20Model%20Driven%20Topological%20Materials%20Multiagent%20System&searchtype=all&source=header) [💻 Code](https://github.com/search?q=TopoMAS%3A%20Large%20Language%20Model%20Driven%20Topological%20Materials%20Multiagent%20System&type=repositories) *Identifies phases.* (Type: System/Agent, Autonomy: -)
- (2025) **An Agentic Framework for Autonomous Materials Computation** [📝 Paper](https://arxiv.org/search/?query=An%20Agentic%20Framework%20for%20Autonomous%20Materials%20Computation&searchtype=all&source=header) [💻 Code](https://github.com/search?q=An%20Agentic%20Framework%20for%20Autonomous%20Materials%20Computation&type=repositories) *Automates computations.* (Type: System/Agent, Autonomy: -)
- (2025-12-22) **An Agentic Framework for Autonomous Materials Computation (v2)** [📝 Paper](https://arxiv.org/abs/2512.19458) [💻 Code](https://github.com/search?q=Agentic%20Framework%20Autonomous%20Materials%20Computation&type=repositories) *Expert-informed agent automating first-principles computational materials workflows with physically consistent parameter selection.* (Type: System/Agent, Autonomy: L3)
- (2026-01-13) **Autonomous Materials Exploration by Integrating Automated Phase Identification and AI-Assisted Reasoning (SARA)** [📝 Paper](https://arxiv.org/abs/2601.08185) [💻 Code](https://github.com/search?q=SARA%20Autonomous%20Materials&type=repositories) *Robotic thin-film synthesis with automated probabilistic phase labeling and human-in-the-loop active learning.* (Type: System/Agent, Autonomy: L2)
- (2026-02-04) **Automated Extraction of Multicomponent Alloy Data Using Large Language Models** [📝 Paper](https://arxiv.org/abs/2602.04602) [💻 Code](https://github.com/search?q=Automated%20Alloy%20Data%20Extraction%20LLM&type=repositories) *LLM pipeline extracting alloy data from 10,000+ papers, producing databases of 37,711 and 148,069 records.* (Type: System/Agent, Autonomy: L2)
- (2026-02-10) **PeroMAS: A Multi-Agent System of Perovskite Material Discovery** [📝 Paper](https://arxiv.org/abs/2602.13312) [💻 Code](https://github.com/search?q=PeroMAS&type=repositories) *Multi-agent framework for perovskite solar cell discovery with literature retrieval, property prediction, and lab validation.* (Type: System/Agent, Autonomy: L3)
- (2026-02-17) **EAA: Automating Materials Characterization with Vision Language Model Agents** [📝 Paper](https://arxiv.org/abs/2602.15294) [💻 Code](https://github.com/search?q=EAA%20Vision%20Language%20Materials&type=repositories) *VLM agent automating microscopy experiments at synchrotron beamlines via MCP instrument control.* (Type: System/Agent, Autonomy: L3)
- (2026-02-24) **Self-Driving Thin Film Laboratory: Autonomous Epitaxial Synthesis via Real-Time RHEED Analysis** [📝 Paper](https://arxiv.org/abs/2602.20432) [💻 Code](https://github.com/search?q=Self-Driving%20Thin%20Film%20Laboratory&type=repositories) *Closed-loop autonomous thin-film growth using computer vision on RHEED images; >30× reduction in experiments.* (Type: Platform, Autonomy: L3)
- (2026-02-27) **JutulGPT: Agentic Scientific Simulation with Execution-Grounded Model Construction** [📝 Paper](https://arxiv.org/abs/2603.00214) [💻 Code](https://github.com/SINTEF-agentlab/JutulGPT) *Interpret–act–validate loop grounded in Julia reservoir simulator for subsurface simulation from natural language.* (Type: System/Agent, Autonomy: L2)
- (2026-03-10) **From Phase Prediction to Phase Design: A ReAct Agent Framework for High-Entropy Alloy Discovery** [📝 Paper](https://arxiv.org/abs/2603.11068) [💻 Code](https://github.com/search?q=ReAct%20Agent%20High-Entropy%20Alloy&type=repositories) *ReAct LLM agent for inverse HEA design using XGBoost surrogate (94.66% accuracy); 2.4–22.8× closer to experimental manifold.* (Type: System/Agent, Autonomy: L3)
- (2026-03-15) **ScienceClaw + Infinite: Autonomous Agents Coordinating Distributed Discovery** [📝 Paper](https://arxiv.org/abs/2603.14312) [💻 Code](https://github.com/lamm-mit/scienceclaw) *Decentralized agent swarm with 300+ interoperable scientific skills and DAG artifact layer; no central coordinator.* (Type: Platform, Autonomy: L3)
- (2025-10) **LLEMA: Evolutionary Search with LLMs for Multi-Objective Materials Discovery** [📝 Paper](https://arxiv.org/abs/2510.22503) [💻 Code](https://github.com/scientific-discovery/LLEMA) *LLM + chemistry-informed evolutionary operators; 14 realistic tasks across electronics, energy, optics, aerospace. (ICLR 2026.)* (Type: System/Agent, Autonomy: L3)
- (2026-04-03) **MatClaw: An Autonomous Code-First LLM Agent for End-to-End Materials Exploration** [📝 Paper](https://arxiv.org/abs/2604.02688) [💻 Code](https://github.com/search?q=MatClaw&type=repositories) *Code-first agent writing Python for multi-step HPC workflows; four-layer memory architecture with ~99% API-call accuracy.* (Type: System/Agent, Autonomy: L3)
- (2026-04-13) **Agentic LLM Reasoning in a Self-Driving Laboratory for Air-Sensitive Lithium Halide Spinel Conductors (A-Lab GPSS)** [📝 Paper](https://arxiv.org/abs/2604.11957) [💻 Code](https://github.com/search?q=A-Lab%20GPSS%20lithium%20halide%20spinel&type=repositories) *Robotic glovebox SDL combining abductive/inductive agentic reasoning with autonomous synthesis & characterization of 352 Li-halide spinel samples; high-conductivity hit rate rises from 1.33% to 5.33% across the campaign.* (Type: Platform, Autonomy: L3)
- (2026-04-15) **MIND: AI Co-Scientist for Material Research** [📝 Paper](https://arxiv.org/abs/2604.13699) [💻 Code](https://github.com/IMMS-Ewha/MIND) *LLM-driven materials co-scientist that iterates over hypothesis refinement, in-silico experimentation with SevenNet-Omni MLIP, and debate-based validation without manual hand-off between stages.* (Type: System/Agent, Autonomy: L3)
- (2026-04-26) **Agentic Fusion of Large Atomic and Language Models to Accelerate Superconductor Discovery (ElementsClaw)** [📝 Paper](https://arxiv.org/abs/2604.23758) [💻 Code](https://github.com/search?q=ElementsClaw%20superconductor&type=repositories) *Couples a 1B-parameter atomic model with LLM-driven semantic reasoning; identifies 68,000 high-confidence superconductor candidates in 28 GPU hours and experimentally verifies four novel materials.* (Type: System/Agent, Autonomy: L3)
- (2026-05-06) **Experiment-as-Code Labs: A Declarative Stack for AI-Driven Scientific Discovery** [📝 Paper](https://arxiv.org/abs/2605.04375) [💻 Code](https://github.com/search?q=Experiment-as-Code%20Labs&type=repositories) *Declarative experiment configurations compiled to device-level APIs with safety, scheduling, and orchestration; deployed end-to-end on a real autonomous battery laboratory.* (Type: Platform, Autonomy: L3)
- (2026-04-13) **AutoSurrogate: An LLM-Driven Multi-Agent Framework for Autonomous Construction of Deep Learning Surrogate Models in Subsurface Flow** [📝 Paper](https://arxiv.org/abs/2604.11945) [💻 Code](https://github.com/search?q=AutoSurrogate%20Subsurface%20Flow&type=repositories) *Multi-agent framework that lets domain scientists build deployment-ready surrogate models from a single natural-language sentence, autonomously retraining and switching architectures on accuracy failures.* (Type: System/Agent, Autonomy: L3)
- (2026-04-18) **ClimAgent: LLM as Agents for Autonomous Open-ended Climate Science Analysis** [📝 Paper](https://arxiv.org/abs/2604.16922) [💻 Code](https://github.com/usail-hkust/ClimAgent) *Autonomous tool-using climate-science framework for end-to-end modeling and analysis across sub-fields; +40.21 % over base LLMs on the new ClimaBench (5 task categories, 2000–2025).* (Type: System/Agent, Autonomy: L3)

### 6. 🌌 Agentic Physics and Astronomy Research
- (2025) **MoRA: Improving physics reasoning in large language models using mixture of refinement agents** [📝 Paper](https://arxiv.org/search/?query=MoRA%3A%20Improving%20physics%20reasoning%20in%20large%20language%20models%20using%20mixture%20of%20refinement%20agents&searchtype=all&source=header) [💻 Code](https://github.com/search?q=MoRA%3A%20Improving%20physics%20reasoning%20in%20large%20language%20models%20using%20mixture%20of%20refinement%20agents&type=repositories) *Improves reasoning.* (Type: System/Agent, Autonomy: -)
- (2025) **LP-COMDA: Physics-informed LLM-agent for automated modulation design in power electronics systems** [📝 Paper](https://arxiv.org/search/?query=LP-COMDA%3A%20Physics-informed%20LLM-agent%20for%20automated%20modulation%20design%20in%20power%20electronics%20systems&searchtype=all&source=header) [💻 Code](https://github.com/search?q=LP-COMDA%3A%20Physics-informed%20LLM-agent%20for%20automated%20modulation%20design%20in%20power%20electronics%20systems&type=repositories) *Designs modulation.* (Type: System/Agent, Autonomy: -)
- (2025) **LLMSat: A large language model-based goal-oriented agent for autonomous space exploration** [📝 Paper](https://arxiv.org/search/?query=LLMSat%3A%20A%20large%20language%20model-based%20goal-oriented%20agent%20for%20autonomous%20space%20exploration&searchtype=all&source=header) [💻 Code](https://github.com/search?q=LLMSat%3A%20A%20large%20language%20model-based%20goal-oriented%20agent%20for%20autonomous%20space%20exploration&type=repositories) *Space exploration.* (Type: System/Agent, Autonomy: -)
- (2025) **ArgoLOOM: Agentic AI for Fundamental Physics from Quarks to Cosmos** [📝 Paper](https://arxiv.org/search/?query=ArgoLOOM%3A%20Agentic%20AI%20for%20Fundamental%20Physics%20from%20Quarks%20to%20Cosmos&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ArgoLOOM%3A%20Agentic%20AI%20for%20Fundamental%20Physics%20from%20Quarks%20to%20Cosmos&type=repositories) *Fundamental physics.* (Type: System/Agent, Autonomy: -)
- (2025) **Agentic AI for Multi-Stage Physics Experiments at a Large-Scale User Facility Particle Accelerator** [📝 Paper](https://arxiv.org/search/?query=Agentic%20AI%20for%20Multi-Stage%20Physics%20Experiments%20at%20a%20Large-Scale%20User%20Facility%20Particle%20Accelerator&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Agentic%20AI%20for%20Multi-Stage%20Physics%20Experiments%20at%20a%20Large-Scale%20User%20Facility%20Particle%20Accelerator&type=repositories) *Experiments at accelerators.* (Type: System/Agent, Autonomy: -)
- (2025) **CosmoAgent: What if LLMs have different world views: Simulating alien civilizations with LLM-based agents** [📝 Paper](https://arxiv.org/search/?query=CosmoAgent%3A%20What%20if%20LLMs%20have%20different%20world%20views%3A%20Simulating%20alien%20civilizations%20with%20LLM-based%20agents&searchtype=all&source=header) [💻 Code](https://github.com/search?q=CosmoAgent%3A%20What%20if%20LLMs%20have%20different%20world%20views%3A%20Simulating%20alien%20civilizations%20with%20LLM-based%20agents&type=repositories) *Simulates civilizations.* (Type: System/Agent, Autonomy: -)
- (2025) **StarWhisper: Agent-based observation assistant system to approach AI astrophysicist** [📝 Paper](https://arxiv.org/search/?query=StarWhisper%3A%20Agent-based%20observation%20assistant%20system%20to%20approach%20AI%20astrophysicist&searchtype=all&source=header) [💻 Code](https://github.com/search?q=StarWhisper%3A%20Agent-based%20observation%20assistant%20system%20to%20approach%20AI%20astrophysicist&type=repositories) *Astrophysics assistant.* (Type: System/Agent, Autonomy: -)
- (2025) **mephisto: Interpreting multi-band galaxy observations with large language model-based agents** [📝 Paper](https://arxiv.org/search/?query=mephisto%3A%20Interpreting%20multi-band%20galaxy%20observations%20with%20large%20language%20model-based%20agents&searchtype=all&source=header) [💻 Code](https://github.com/search?q=mephisto%3A%20Interpreting%20multi-band%20galaxy%20observations%20with%20large%20language%20model-based%20agents&type=repositories) *Galaxy observations.* (Type: System/Agent, Autonomy: -)
- (2025) **AI Agents for ground-based gamma astronomy** [📝 Paper](https://arxiv.org/search/?query=AI%20Agents%20for%20ground-based%20gamma%20astronomy&searchtype=all&source=header) [💻 Code](https://github.com/search?q=AI%20Agents%20for%20ground-based%20gamma%20astronomy&type=repositories) *Gamma astronomy.* (Type: System/Agent, Autonomy: -)
- (2025) **The AI Cosmologist I: An Agentic System for Automated Data Analysis** [📝 Paper](https://arxiv.org/search/?query=The%20AI%20Cosmologist%20I%3A%20An%20Agentic%20System%20for%20Automated%20Data%20Analysis&searchtype=all&source=header) [💻 Code](https://github.com/search?q=The%20AI%20Cosmologist%20I%3A%20An%20Agentic%20System%20for%20Automated%20Data%20Analysis&type=repositories) *Automated analysis.* (Type: System/Agent, Autonomy: -)
- (2025) **OpenFOAMGPT: Fluid Dynamics GPT** [📝 Paper](https://arxiv.org/search/?query=OpenFOAMGPT%3A%20Fluid%20Dynamics%20GPT&searchtype=all&source=header) [💻 Code](https://github.com/search?q=OpenFOAMGPT%3A%20Fluid%20Dynamics%20GPT&type=repositories) *Automates dynamics.* (Type: Paper, Autonomy: -)
- (2025) **OpenFOAMGPT 2.0: Updated Version** [📝 Paper](https://arxiv.org/search/?query=OpenFOAMGPT%202.0%3A%20Updated%20Version&searchtype=all&source=header) [💻 Code](https://github.com/search?q=OpenFOAMGPT%202.0%3A%20Updated%20Version&type=repositories) *Enhanced dynamics.* (Type: Paper, Autonomy: -)
- (2025) **LLM-Agent: LLM for Mechanics** [📝 Paper](https://arxiv.org/search/?query=LLM-Agent%3A%20LLM%20for%20Mechanics&searchtype=all&source=header) [💻 Code](https://github.com/search?q=LLM-Agent%3A%20LLM%20for%20Mechanics&type=repositories) *Mechanics agent.* (Type: System/Agent, Autonomy: -)
- (2025) **MechAgents: Computational Solid Mechanics** [📝 Paper](https://arxiv.org/search/?query=MechAgents%3A%20Computational%20Solid%20Mechanics&searchtype=all&source=header) [💻 Code](https://github.com/search?q=MechAgents%3A%20Computational%20Solid%20Mechanics&type=repositories) *Mechanics simulations (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025) **AutoGen-FEM: Finite Element** [📝 Paper](https://arxiv.org/search/?query=AutoGen-FEM%3A%20Finite%20Element&searchtype=all&source=header) [💻 Code](https://github.com/search?q=AutoGen-FEM%3A%20Finite%20Element&type=repositories) *Finite element analysis.* (Type: Paper, Autonomy: -)
- (2025) **k-agents: Quantum Processor Control** [📝 Paper](https://arxiv.org/search/?query=k-agents%3A%20Quantum%20Processor%20Control&searchtype=all&source=header) [💻 Code](https://github.com/search?q=k-agents%3A%20Quantum%20Processor%20Control&type=repositories) *Controls processors (Level 2).* (Type: System/Agent, Autonomy: L2)
- (2025-12-08) **Automating High Energy Physics Data Analysis with LLM-Powered Agents** [📝 Paper](https://arxiv.org/abs/2512.07785) [💻 Code](https://huggingface.co/HWresearch/LLM4HEP) *Supervisor-coder agent with Snakemake for autonomous HEP analysis on ATLAS Higgs data. (NeurIPS 2025 ML&PS.)* (Type: System/Agent, Autonomy: L3)
- (2025-12-31) **AstroReview: An LLM-driven Multi-Agent Framework for Telescope Proposal Peer Review** [📝 Paper](https://arxiv.org/abs/2512.24754) [💻 Code](https://github.com/search?q=AstroReview&type=repositories) *Three-stage multi-agent framework for automating telescope proposal review; 87% acceptance prediction accuracy.* (Type: System/Agent, Autonomy: L2)
- (2026-01-31) **GRACE: An Agentic AI for Particle Physics Experiment Design and Simulation** [📝 Paper](https://arxiv.org/abs/2602.15039) [💻 Code](https://github.com/just5034/GRACE_whitepaper_data) *Simulation-native agent building toy Monte Carlo simulations and proposing detector geometry modifications.* (Type: System/Agent, Autonomy: L3)
- (2026-02-10) **Talking with the Latents: How to Convert Your LLM into an Astronomer** [📝 Paper](https://arxiv.org/abs/2602.09670) [💻 Code](https://github.com/search?q=Talking%20with%20Latents%20Astronomer&type=repositories) *Teacher-student distillation fusing physical latent features into LLMs via adapters; outperforms Gemini 3 Pro.* (Type: System/Agent, Autonomy: -)
- (2026-03-05) **Agentic AI–Physicist Collaboration in Experimental Particle Physics: LEP Open Data** [📝 Paper](https://arxiv.org/abs/2603.05735) [💻 Code](https://github.com/search?q=Agentic%20AI%20Physicist%20LEP&type=repositories) *AI agents conducted a full thrust-distribution measurement on ALEPH data under physicist supervision.* (Type: System/Agent, Autonomy: L2)
- (2026-03-20) **AI Agents Can Already Autonomously Perform Experimental High Energy Physics** [📝 Paper](https://arxiv.org/abs/2603.20179) [💻 Code](https://github.com/search?q=AI%20Agents%20Experimental%20HEP&type=repositories) *Claude autonomously executing complete HEP analysis pipelines on open ALEPH, DELPHI, and CMS datasets.* (Type: System/Agent, Autonomy: L3)
- (2026-03-23) **The FERMIACC: Agents for Particle Theory** [📝 Paper](https://arxiv.org/abs/2603.22538) [💻 Code](https://github.com/search?q=FERMIACC&type=repositories) *Scaffolded agent system autonomously generating and validating QFT hypotheses via adversarial proposer-critic loop.* (Type: System/Agent, Autonomy: L3)
- (2026-03-27) **Agentic Diagrammatica: Towards Autonomous Symbolic Computation in HEP** [📝 Paper](https://arxiv.org/abs/2603.26990) [💻 Code](https://github.com/tonymenzo/heptapod) *HEPTAPOD framework with Diagrammatica module for exact tree-level FeynCalc calculations.* (Type: System/Agent, Autonomy: L2)
- (2026-03-30) **ALBERT: Autonomous Discovery of Particle Physics Theories from Experimental Data** [📝 Paper](https://arxiv.org/abs/2603.28935) [💻 Code](https://github.com/search?q=ALBERT%20Particle%20Physics%20Theory&type=repositories) *Neuro-symbolic RL transformer that rediscovered the Standard Model and predicted top quark mass (178.9±5.0 GeV) from LEP data.* (Type: System/Agent, Autonomy: L3)
- (2026-04-04) **LensAgent: A Self-Evolving Agent for Autonomous Gravitational Lensing Inference** [📝 Paper](https://arxiv.org/abs/2604.03691) [💻 Code](https://github.com/search?q=LensAgent&type=repositories) *First training-free LLM-driven agent for autonomous strong gravitational lens modeling and dark matter substructure detection.* (Type: System/Agent, Autonomy: L3)
- (2026-04-09) **A Model Context Protocol Server for Quantum Execution in Hybrid Quantum-HPC Environments** [📝 Paper](https://arxiv.org/abs/2604.08318) [💻 Code](https://github.com/search?q=MCP%20Quantum%20Execution%20HPC&type=repositories) *MCP server enabling LLM agents to autonomously orchestrate OpenQASM execution and asynchronous QPU/HPC pipelines via CUDA-Q and Quantinuum on the ABCI-Q platform.* (Type: Platform, Autonomy: L2)
- (2026-04-16) **PRL-Bench: A Comprehensive Benchmark Evaluating LLMs' Capabilities in Frontier Physics Research** [📝 Paper](https://arxiv.org/abs/2604.15411) [💻 Code](https://github.com/search?q=PRL-Bench%20Physics&type=repositories) *End-to-end physics research benchmark built from 100 PRL papers across astrophysics, condensed matter, HEP, quantum information, and statistical physics; best overall LLM score below 50.* (Type: Benchmark/Eval, Autonomy: -)
- (2026-04-24) **Dr.Sai: An Agentic AI for Real-World Physics Analysis at BESIII** [📝 Paper](https://arxiv.org/abs/2604.22541) [💻 Code](https://github.com/search?q=Dr.Sai%20BESIII&type=repositories) *Multi-agent system translating natural-language requests into rigorous BESIII analysis workflows; autonomously re-measured ten J/ψ decay branching fractions in the live IHEP computing environment.* (Type: System/Agent, Autonomy: L3)
- (2026-05-01) **Towards Multi-Agent Autonomous Reasoning in Hydrodynamics** [📝 Paper](https://arxiv.org/abs/2605.01102) [💻 Code](https://github.com/search?q=Multi-Agent%20Hydrodynamics%20Reasoning&type=repositories) *Layer-Execution-Graph multi-agent system for hydrodynamics that tackles single-agent context saturation; 93.6 % factual precision and 100 % pass rate on a 37-query benchmark.* (Type: System/Agent, Autonomy: L2)
- (2026-05-07) **AI CFD Scientist: Toward Open-Ended Computational Fluid Dynamics Discovery with Physics-Aware AI Agents** [📝 Paper](https://arxiv.org/abs/2605.06607) [💻 Code](https://github.com/search?q=AI%20CFD%20Scientist&type=repositories) *Physics-aware autonomous CFD agent spanning literature ideation, validated solver execution, and a vision-language gate that catches 14/16 planted physics failures missed by baselines; discovers a turbulence-model improvement reducing benchmark error ~8 %.* (Type: System/Agent, Autonomy: L3)

---

## 🔄 Workflow Stages
1. **Hypothesis Generation**: literature mining.
2. **Experimental Planning**: tool orchestration.
3. **Data Analysis**: validation.
4. **Synthesis**: refinement.

## 🏥 Domain Applications

#### 🧬 Life Sciences
- **Hypothesis**: [CellVoyager](https://arxiv.org/search/?query=CellVoyager&searchtype=all&source=header) [💻 Code](https://github.com/search?q=CellVoyager&type=repositories) — *Novel links* (Autonomy: Level 3)
- **Design**: [CRISPR-GPT](https://www.biorxiv.org/content/10.1101/2024.04.25.591003v1) [💻 Code](https://github.com/cong-lab/crispr-gpt-pub) — *Plans* (Autonomy: Level 2-3)
- **Analysis**: [SpatialAgent](https://www.biorxiv.org/content/10.1101/2025.04.03.646459v1) [💻 Code](https://github.com/Genentech/SpatialAgent) — *Integration* (Autonomy: Level 2-3)
- **Validation**: [The virtual lab](https://www.biorxiv.org/content/10.1101/2024.11.11.623744v1) [💻 Code](https://github.com/zou-group/virtual-lab) — *Confirmations* (Autonomy: Level 3)

#### 🧪 Chemistry
- **Optimization**: [Coscientist](https://doi.org/10.1038/s41586-023-06792-0) [💻 Code](https://github.com/gomesgroup/coscientist) — *Cross-coupling* (Autonomy: Level 3)
- **Design**: [OSDA Agent](https://arxiv.org/search/?query=OSDA%20Agent&searchtype=all&source=header) [💻 Code](https://github.com/search?q=OSDA%20Agent&type=repositories) — *Zeolites* (Autonomy: Level 3)
- **Mining**: [ChemMiner](https://arxiv.org/search/?query=ChemMiner&searchtype=all&source=header) [💻 Code](https://github.com/search?q=ChemMiner&type=repositories) — *Datasets* (Autonomy: Level 2)
- **Workflows**: [El Agente Q](https://arxiv.org/search/?query=El%20Agente%20Q&searchtype=all&source=header) [💻 Code](https://github.com/search?q=El%20Agente%20Q&type=repositories) — *Predictions* (Autonomy: Level 2)

#### 💎 Materials Science
- **Design**: [AtomAgents](https://arxiv.org/search/?query=AtomAgents&searchtype=all&source=header) [💻 Code](https://github.com/search?q=AtomAgents&type=repositories) — *Alloys* (Autonomy: Level 3)
- **Simulation**: [Foam-Agent](https://arxiv.org/search/?query=Foam-Agent&searchtype=all&source=header) [💻 Code](https://github.com/search?q=Foam-Agent&type=repositories) — *CFD* (Autonomy: Level 2)
- **Platforms**: [LLMatDesign](https://arxiv.org/search/?query=LLMatDesign&searchtype=all&source=header) [💻 Code](https://github.com/search?q=LLMatDesign&type=repositories) — *Inverse design* (Autonomy: Level 3)
- **Air-Sensitive SDL**: [A-Lab GPSS](https://arxiv.org/abs/2604.11957) — *Robotic synthesis of Li-halide spinel conductors* (Autonomy: Level 3)
- **Superconductor Discovery**: [ElementsClaw](https://arxiv.org/abs/2604.23758) — *68k candidates in 28 GPU h, 4 experimentally verified* (Autonomy: Level 3)
- **Declarative SDL Stack**: [Experiment-as-Code Labs](https://arxiv.org/abs/2605.04375) — *Compiles experiments to device APIs* (Autonomy: Level 3)

#### 🌌 Physics and Astronomy
- **Interpretation**: [CosmoAgent](https://arxiv.org/search/?query=CosmoAgent&searchtype=all&source=header) [💻 Code](https://github.com/search?q=CosmoAgent&type=repositories) — *Insights* (Autonomy: Level 2-3)
- **Automation**: [OpenFOAMGPT](https://arxiv.org/search/?query=OpenFOAMGPT&searchtype=all&source=header) [💻 Code](https://github.com/search?q=OpenFOAMGPT&type=repositories) — *Analyses* (Autonomy: Level 2)
- **Control**: [k-agents](https://arxiv.org/search/?query=k-agents&searchtype=all&source=header) [💻 Code](https://github.com/search?q=k-agents&type=repositories) — *Optimization* (Autonomy: Level 2)
- **Theory Discovery**: [ALBERT](https://arxiv.org/abs/2603.28935) — *Rediscovered Standard Model from LEP data* (Autonomy: Level 3)
- **HEP Analysis**: [AI Agents in Experimental HEP](https://arxiv.org/abs/2603.20179) — *End-to-end autonomous HEP pipeline* (Autonomy: Level 3)
- **Lensing**: [LensAgent](https://arxiv.org/abs/2604.03691) — *Autonomous gravitational lens modeling* (Autonomy: Level 3)
- **BESIII Analysis**: [Dr.Sai](https://arxiv.org/abs/2604.22541) — *Autonomously re-measured ten J/ψ branching fractions* (Autonomy: Level 3)
- **CFD Discovery**: [AI CFD Scientist](https://arxiv.org/abs/2605.06607) — *Vision-gated autonomous turbulence-model improvement* (Autonomy: Level 3)

#### 🧮 Mathematics
- **Theorem Exploration**: [AI Co-Mathematician](https://arxiv.org/abs/2605.06651) — *Stateful agentic workbench, 48% on FrontierMath Tier 4* (Autonomy: -)
- **Open Erdős Problems**: [Aletheia](https://arxiv.org/abs/2602.10177) — *Solved four open Erdős conjectures* (Autonomy: -)
- **Formalization**: [Sphere Packing in 8D](https://arxiv.org/abs/2604.23468) — *Lean formalization of Viazovska's proof, finished by autoformalization model "Gauss"* (Autonomy: -)
- **Human-AI Workflow**: [Intentmaking & Sensemaking with AlphaEvolve](https://arxiv.org/abs/2605.05921) — *Empirical study of how mathematicians collaborate with AlphaEvolve* (Autonomy: -)

#### 🌍 Earth & Climate Science
- **Climate Analysis**: [ClimAgent](https://arxiv.org/abs/2604.16922) — *Autonomous open-ended climate science, +40 % on ClimaBench* (Autonomy: Level 3)
- **Subsurface Surrogates**: [AutoSurrogate](https://arxiv.org/abs/2604.11945) — *NL-to-deployment surrogate models for subsurface flow* (Autonomy: Level 3)
- **Geosciences**: [PANGAEA-GPT](https://arxiv.org/abs/2602.21351) — *Hierarchical multi-agent system over six geoscientific data archives* (Autonomy: -)
- **Hydrodynamics**: [Multi-Agent Hydrodynamics Reasoning](https://arxiv.org/abs/2605.01102) — *Layer-Execution-Graph multi-agent reasoning, 93.6 % factual precision* (Autonomy: Level 2)

#### 🧠 Agentic Failure Modes & Safety
- **Astrophysics**: [Plausible but Wrong](https://arxiv.org/abs/2604.25345) — *Silent generation of physically incorrect results in CMBAgent across 18 tasks*
- **Biosecurity**: [BioVeil MATRIX](https://arxiv.org/abs/2605.00927) — *Defensive taxonomy with 10 categories, 22 techniques for bio-agent red-teaming*

### Challenges and Directions
Benchmarks like DiscoveryWorld for reproducibility. Ethics: equitable access. Future: Nobel-Turing Test, global networks.

---

## 🔑 Key Citations
- [From AI for Science to Agentic Science: A Survey on Autonomous Scientific Discovery](https://arxiv.org/abs/2508.14111) (2025-08-18)
- [Agentic AI for Scientific Discovery: A Survey of Progress, Challenges, and Future Directions](https://arxiv.org/abs/2503.08979) (2025-03-12)
- [The (R)evolution of Scientific Workflows in the Agentic AI Era: Towards Autonomous Science](https://arxiv.org/abs/2509.09915) (2025-09-12)
- [The AI scientist: Towards fully automated open-ended scientific discovery](https://arxiv.org/abs/2408.06292) (2024-08-12; **Nature 2026**) — [GitHub](https://github.com/SakanaAI/AI-Scientist)
- [The AI Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Search](https://arxiv.org/abs/2504.08066) (2025-04-10) — [GitHub](https://github.com/SakanaAI/AI-Scientist-v2)
- [Kosmos: An AI Scientist for Autonomous Discovery](https://arxiv.org/abs/2511.02824) (2025-11-04)
- [SR-Scientist: Scientific Equation Discovery With Agentic AI](https://arxiv.org/abs/2510.11661) (2025-10-13)
- [The Denario project: Deep knowledge AI agents for scientific discovery](https://arxiv.org/abs/2510.26887) (2025-10-30) — [GitHub](https://github.com/denario-sw/denario)
- [AgenticSciML: Collaborative Multi-Agent Systems for Emergent Discovery in Scientific Machine Learning](https://arxiv.org/abs/2511.07262) (2025-11-10)
- [Robin: A multi-agent system for automating scientific discovery](https://arxiv.org/abs/2505.13400) (2025-05-19)
- [Enhancing Agentic Autonomous Scientific Discovery with Vision-Language Model Capabilities](https://arxiv.org/abs/2511.14631) (2025-11-18) — [GitHub](https://github.com/CMBAgents/cmbagent)
- [Autonomous chemical research with large language models](https://doi.org/10.1038/s41586-023-06792-0) (2023-12-20) — [GitHub](https://github.com/gomesgroup/coscientist)
- [DiscoveryWorld: A virtual environment for developing and evaluating automated scientific discovery agents](https://arxiv.org/abs/2406.06769) (2024-06-10) — [GitHub](https://github.com/allenai/discoveryworld)
- [Biomni: A general-purpose biomedical AI agent](https://www.biorxiv.org/content/10.1101/2025.05.30.656746v1) (2025-05-30) — [GitHub](https://github.com/snap-stanford/Biomni)
- [SpatialAgent: An autonomous AI agent for spatial biology](https://www.biorxiv.org/content/10.1101/2025.04.03.646459v1) (2025-04-03) — [GitHub](https://github.com/Genentech/SpatialAgent)
- [MOSAIC: Collective Intelligence for AI-Assisted Chemical Synthesis](https://www.nature.com/articles/s41586-026-10131-4) (2026) — [GitHub](https://github.com/haoteli/MOSAIC)
- [EvoScientist: Multi-Agent Evolving AI Scientists](https://arxiv.org/abs/2603.08127) (2026-03-09) — [GitHub](https://github.com/EvoScientist/EvoScientist)
- [CORAL: Autonomous Multi-Agent Evolution for Open-Ended Discovery](https://arxiv.org/abs/2604.01658) (2026-04-01) — [GitHub](https://github.com/Human-Agent-Society/CORAL)
- [BioMedAgent: Self-evolving multi-agent for biomedical data analysis](https://www.nature.com/articles/s41551-026-01634-6) (2026) — [GitHub](https://github.com/BOBQWERA/BioMedAgent)
- [ALBERT: Autonomous Discovery of Particle Physics Theories](https://arxiv.org/abs/2603.28935) (2026-03-30)
- [FermiLink: Multidomain Autonomous Scientific Simulations](https://arxiv.org/abs/2604.03460) (2026-04-03) — [GitHub](https://github.com/TaoELi/FermiLink)
- [HeurekaBench: A Benchmarking Framework for AI Co-scientist](https://arxiv.org/abs/2601.01678) (2026-01-04; **ICLR 2026**) — [GitHub](https://github.com/mlbio-epfl/HeurekaBench)
- [EvoMaster: A Foundational Evolving Agent Framework for Agentic Science at Scale](https://arxiv.org/abs/2604.17406) (2026-04-19) — [GitHub](https://github.com/sjtu-sai-agents/EvoMaster)
- [Agentic LLM Reasoning in a Self-Driving Laboratory for Air-Sensitive Lithium Halide Spinel Conductors (A-Lab GPSS)](https://arxiv.org/abs/2604.11957) (2026-04-13)
- [Dr.Sai: An Agentic AI for Real-World Physics Analysis at BESIII](https://arxiv.org/abs/2604.22541) (2026-04-24)
- [AI Co-Mathematician: Accelerating Mathematicians with Agentic AI](https://arxiv.org/abs/2605.06651) (2026-05-07)
- [LABBench2: An Improved Benchmark for AI Systems Performing Biology Research](https://arxiv.org/abs/2604.09554) (2026-04-14) — [GitHub](https://github.com/EdisonScientific/labbench2)
- [Toward Autonomous Long-Horizon Engineering for ML Research (AiScientist)](https://arxiv.org/abs/2604.13018) (2026-04-14) — [GitHub](https://github.com/AweAI-Team/AiScientist)
- [Plausible but Wrong: Agentic Failures in Astrophysical Workflows](https://arxiv.org/abs/2604.25345) (2026-04-28)
- [A Milestone in Formalization: The Sphere Packing Problem in Dimension 8](https://arxiv.org/abs/2604.23468) (2026-04-25)
- [ClimAgent: LLM as Agents for Autonomous Open-ended Climate Science Analysis](https://arxiv.org/abs/2604.16922) (2026-04-18) — [GitHub](https://github.com/usail-hkust/ClimAgent)
- [Camyla: Scaling Autonomous Research in Medical Image Segmentation](https://arxiv.org/abs/2604.10696) (2026-04-12)

## 🔗 Related Awesome Lists / Repos
- [AgenticScience/Awesome-Agent-Scientists](https://github.com/AgenticScience/Awesome-Agent-Scientists) — paper list of agents for science
- [skyworkai/Awesome-AI-Scientist](https://github.com/skyworkai/Awesome-AI-Scientist) — papers, code, datasets for AI scientist
- [XiaoYuan98/Awesome-Human-Motion-Video-Generation](https://github.com/XiaoYuan98/Awesome-Human-Motion-Video-Generation)
- [jiajunhua/awesome-transformer-vl](https://github.com/jiajunhua/awesome-transformer-vl)
- [wuzhipeng/awesome-ai-agent-in-education](https://github.com/wuzhipeng/awesome-ai-agent-in-education)
- [ashishpatel26/LLM-Agents-Papers](https://github.com/ashishpatel26/LLM-Agents-Papers)
- [fuzhen-yu/LLM4DC](https://github.com/fuzhen-yu/LLM4DC)
- [e2b-dev/awesome-ai-agents](https://github.com/e2b-dev/awesome-ai-agents)
- [tmgthb/Autonomous-Agents-in-Biomedicine-and-Chemistry](https://github.com/tmgthb/Autonomous-Agents-in-Biomedicine-and-Chemistry)
- [Tengfei-Wang/Awesome-AI-Agent-for-Materials](https://github.com/Tengfei-Wang/Awesome-AI-Agent-for-Materials)
- [probertson91/awesome-scientific-writing](https://github.com/probertson91/awesome-scientific-writing)
- [RManLuo/Awesome-LLM-KG](https://github.com/RManLuo/Awesome-LLM-KG)

## 📈 Star History
[![Star History Chart](https://api.star-history.com/svg?repos=wu-yc/Awesome-Agentic-Labs&type=date&legend=top-left)](https://www.star-history.com/#wu-yc/Awesome-Agentic-Labs&type=date&legend=top-left)

---

## 🤝 Contributing
We warmly welcome contributions via pull request.

When adding/editing an entry, please try to include:
- **Exact date** (arXiv “Submitted on …” / bioRxiv “Posted …” / publisher online date)
- **Canonical paper link**
- **Official code repo** (or a note “code not released”)

### 📚 Citation
If you find this repository useful, please consider citing:
```bibtex
@misc{awesome-agentic-labs,
  author = {Awesome Agentic Labs Contributors},
  title = {Awesome Agentic Labs: The AI Infrastructure of Autonomous Science},
  year = {2026},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/wu-yc/Awesome-Agentic-Labs}},
}
```

<div align="center">
 Star us on GitHub! 🌟
</div>

Awesome-Agentic-Labs

File Explorer

Today's picks

Was this content helpful?

Recent Feedback

Awesome-Agentic-Labs

File Explorer

Today's picks

Was this content helpful?

Recent Feedback

Download README