Repolode
Submit
EvoMap

EvoMap/AutoResearch

Open-source agent workflow that takes an AI/ML research idea through experiment planning, code, execution, and independent review into an evidence package for paper writing.

View on GitHub
Stars
3,091
Forks
278
Open issues
6
Last commit
2 months ago

What is it?

AutoResearch is an agent workflow for AI and machine learning research. It can start from a research idea the user provides, or generate candidate directions from recent papers, community discussions, and open-source trends. From there it moves through experiment planning, implementation, review, execution, result analysis, and independent evaluation.

The pipeline is stateful and recoverable. Plans, code, run logs, metrics, failure causes, critic reports, and blind reviews are written to disk, so researchers can inspect the process, take it over, or stop it. Idea generation uses at least three distinct models for cross-review, and a pilot run tests feasibility before full experiments begin.

The project is configured through a single local provider file that maps roles to one or more compatible model endpoints. Its execution stage runs on the Claude Code CLI with a Bun-based runtime, and it supports user-written knowledge base files that feed into idea generation.

Key features

  • 01Generates research directions from recent papers and community signals, or starts from a user-provided idea.
  • 02Uses at least three distinct models for cross-review during idea generation.
  • 03Persists plans, code, queues, logs, and conclusions so long experiments can resume after interruption.
  • 04Runs a low-cost pilot before scaling to full experiments.
  • 05Records critic reports and blind reviews, and preserves negative results.

Use cases

  • Turning a vague AI research interest into a reviewed, testable experiment plan.
  • Running long machine learning experiments that need to resume after interruptions.
  • Assembling a traceable evidence package before writing a paper.
  • Auditing how an automated research process reached its conclusions.

Quick start

  1. 1

    Clone the repository

    git clone https://github.com/EvoMap/AutoResearch.git
  2. 2

    Enter the project directory and run the environment bringup script

    cd AutoResearch && bash scripts/bringup.sh
  3. 3

    Create local config files, add API keys and model settings, then test the APIs

  4. 4

    Generate research ideas with the default entrypoint

    .venv/bin/python idea_generation.py

Things to know

  • Requires API keys for model services and may incur API charges.
  • Idea execution additionally needs the official Claude Code CLI, Bun, and the Ralph Loop plugin.
  • Experiments may require GPU, data, and disk resources that you must provide.
  • The README states the system cannot guarantee that every conclusion is correct.

In collections

Open-source AI agent frameworks & apps

Topics

#ai-research#llm#agents#machine-learning#experiment-automation#multi-model#paper-writing

FAQ

What is AutoResearch?
Open-source agent workflow that takes an AI/ML research idea through experiment planning, code, execution, and independent review into an evidence package for paper writing. AutoResearch is an agent workflow for AI and machine learning research. It can start from a research idea the user provides, or generate candidate directions from recent papers, community discussions, and open-source trends. From there it moves through experiment planning, implementation, review, execution, result analysis, and independent evaluation.
Is AutoResearch free to use? What license does it have?
AutoResearch is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is AutoResearch still maintained?
Updated within 6 months. The last commit to AutoResearch was on Sep 16, 2026.
What can I use AutoResearch for?
Turning a vague AI research interest into a reviewed, testable experiment plan. Running long machine learning experiments that need to resume after interruptions. Assembling a traceable evidence package before writing a paper. Auditing how an automated research process reached its conclusions.
How do I install AutoResearch?
Clone the repository: git clone https://github.com/EvoMap/AutoResearch.git Enter the project directory and run the environment bringup script: cd AutoResearch && bash scripts/bringup.sh Create local config files, add API keys and model settings, then test the APIs Generate research ideas with the default entrypoint: .venv/bin/python idea_generation.py
Where can I get AutoResearch?
The source code and installation instructions are on GitHub: https://github.com/EvoMap/AutoResearch

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects