Survey of notable research agent projects from GitHub and arXiv
Python
0
7 commits
updated Feb 18, 2026
README
sandbox-research-agent-survey
A survey of notable research agent projects from GitHub and arXiv, covering autonomous web research, scientific discovery, literature synthesis, multi-agent frameworks, and evaluation benchmarks.
📂 Survey
Head to survey/ for the full survey, including:
- 10 detailed project write-ups — each in its own folder with architecture diagrams, feature lists, and references
- Summary table — quick comparison across all projects
- Taxonomy — organized by approach, output type, and maturity
- Key trends — what's happening in the research agent space (2024–2025)
Projects Covered
| # | Project | One-liner |
|---|---|---|
| 1 | GPT-Researcher | Autonomous deep research agent with plan-and-execute architecture |
| 2 | STORM | Wikipedia-article generation via multi-perspective questioning (Stanford) |
| 3 | OpenScholar | RAG over 45M papers for scientific literature synthesis (AI2, Nature) |
| 4 | The AI Scientist | Idea → code → experiment → paper, fully automated (Sakana AI) |
| 5 | Agent Laboratory | Multi-role lab simulation with co-pilot mode |
| 6 | PaperQA2 | Superhuman scientific QA with agentic RAG (Future House) |
| 7 | Alibaba DeepResearch | Multi-turn web research with cross-checking (Tongyi Lab) |
| 8 | ScienceAgentBench | Benchmark: 102 real scientific tasks, best agent solves ~32% |
| 9 | awesome-deep-research-agent | Curated paper list mapping the deep research landscape |
| 10 | AutoGen | Multi-agent conversation framework by Microsoft (52k+ ⭐) |
License
This survey is provided for informational purposes. Each referenced project has its own license.
Contributors
7 commits
Similar repos
Languages
Python
100.0%
Survey of notable research agent projects from GitHub and arXiv
Python
0
7 commits
updated Feb 18, 2026
README
sandbox-research-agent-survey
A survey of notable research agent projects from GitHub and arXiv, covering autonomous web research, scientific discovery, literature synthesis, multi-agent frameworks, and evaluation benchmarks.
📂 Survey
Head to survey/ for the full survey, including:
- 10 detailed project write-ups — each in its own folder with architecture diagrams, feature lists, and references
- Summary table — quick comparison across all projects
- Taxonomy — organized by approach, output type, and maturity
- Key trends — what's happening in the research agent space (2024–2025)
Projects Covered
| # | Project | One-liner |
|---|---|---|
| 1 | GPT-Researcher | Autonomous deep research agent with plan-and-execute architecture |
| 2 | STORM | Wikipedia-article generation via multi-perspective questioning (Stanford) |
| 3 | OpenScholar | RAG over 45M papers for scientific literature synthesis (AI2, Nature) |
| 4 | The AI Scientist | Idea → code → experiment → paper, fully automated (Sakana AI) |
| 5 | Agent Laboratory | Multi-role lab simulation with co-pilot mode |
| 6 | PaperQA2 | Superhuman scientific QA with agentic RAG (Future House) |
| 7 | Alibaba DeepResearch | Multi-turn web research with cross-checking (Tongyi Lab) |
| 8 | ScienceAgentBench | Benchmark: 102 real scientific tasks, best agent solves ~32% |
| 9 | awesome-deep-research-agent | Curated paper list mapping the deep research landscape |
| 10 | AutoGen | Multi-agent conversation framework by Microsoft (52k+ ⭐) |
License
This survey is provided for informational purposes. Each referenced project has its own license.
Similar repos
Contributors
7 commits
Languages
Python
100.0%