research-agents

A curated list of autonomous improvement loops, research agents, and autoresearch-style systems inspired by Karpathy's autoresearch.

Group Evolving Agents: Open-Ended Self-Improvement via Experience Sharing

Agentic quant research workbench with CSV data inspection, local Codex and compatible APIs, predictive models, independent reviews, and reproducible backtests.

Learning to Evaluate Before Improving: Automatic Rubric Induction for Automatic Research Agents

IdeaTaste - A benchmark for research taste: let LLM agents propose research ideas, then check how many of them the research area actually went on to do.

LitReview Skill is an installable agent skill for end-to-end literature review generation. It helps agents conduct literature reviews with a well-designed and widely used review framework so the search process is broad, iterative, and less likely to miss relevant articles.

Open-source deep research harness for evidence-grounded reports: multi-agent planning, independent section research, and targeted report revision.

Open-source Bittensor subnet for a global AI agent competition in sales intelligence

SutroYaro — Sutro Group research workspace for energy-efficient AI training. Point any coding agent at the repo and it becomes a research agent. 34 experiments, eval environment, weekly catch-ups, multi-researcher workflow.

🤖 CodeForge AI: An autonomous multi-agent coding system powered by LangGraph for agentic software development and automated workflows. SOTA custom agentic GraphRag, shared-state memory, auto-model routing for cost optimization, and a range of custom tooling.

🔍 DISCOVER — Auditable knowledge graph of claims and contributions from primary literature. See the state of a field, contradictions, and gaps.

Curated paper-related AI skills and GitHub repositories for idea discovery, literature search, experiments, writing, citations, LaTeX/DOCX, review, and submission.

Faraday: An Autonomous Web Research Agent (LangGraph/Streamlit). 🕵️♀️ Investigates queries using dynamic tools (Tavily, Google, NewsAPI, etc.), gathers multi-source info, and synthesizes structured reports in a Streamlit UI. Features agentic workflow & source tracking.

Evaluation code, tasks, prompts, and scientific tools for EarthVerse.

📝 REVIEW — 10-stage pipeline turning a paper PDF into claim/evidence map, novelty and rigor checks, and a self-critique pass. Built to help authors improve.

Continues as JevGPT: github.com/sunyasheng/JevGPT (arXiv 2610.18272)

The official Python SDK for getclean, the data layer for research agents.

The official TypeScript SDK for getclean, the data layer for research agents.

The getclean API contract: the data layer for research agents, in one OpenAPI 3.1 file.

Six MCP servers that automate the full academic research pipeline — from refining a vague research question to generating a publication-ready report. Each server handles a distinct stage of the workflow: question development, data processing, code generation, script execut