whenpoem/aiscientist

8 stars · Last commit 2026-07-20

An AI-driven research workflow for generating ideas, running experiments, and writing scientific papers.

Workflow & Automation TDD & Testing Documentation Design & UI DevOps & Infra Productivity AI & Prompting Data & ML Security

README preview

# ClaudeScientist

**Research memory, result verification, and a local monitoring interface for Claude Code and Codex.**

[![version](https://img.shields.io/badge/version-v5.1.4-blue)](https://github.com/whenpoem/aiscientist/releases) [![python](https://img.shields.io/badge/python-%E2%89%A53.11-3776AB?logo=python&logoColor=white)](https://www.python.org/) [![CI](https://github.com/whenpoem/aiscientist/actions/workflows/ci.yml/badge.svg)](https://github.com/whenpoem/aiscientist/actions/workflows/ci.yml) [![license](https://img.shields.io/badge/license-MIT-yellow)](LICENSE)

> 中文版本: [README.zh-CN.md](README.zh-CN.md)

ClaudeScientist adds persistent research records, result checks, statistical
proof tools, and a terminal monitoring interface to Claude Code and Codex.
It stores hypotheses, evidence, comparisons, experiment records, and user
interventions in a local SQLite database inside each research project.

The agent can use these records to compare hypotheses, run checked experiments,
and verify reported results. A second terminal can display the current research
state and accept user decisions while the agent is working.

**Current version**: v5.1.4 separates public installation, per-workspace
configuration, and source development. It adds `claudescientist configure`,
loads `.research-agent/config.toml` automatically, packages Lean as a disabled

View full repository on GitHub →