researchpad 0.1.0

AI-powered autonomous ML experimentation companion

ResearchPad

AI-powered autonomous ML experimentation companion.

ResearchPad provides a web UI and cursor commands for running iterative ML experiment loops. It tracks experiments, manages research artifacts, performs debug analysis, and surfaces insights -- all driven by AI coding agents.

Quick Start

pip install researchpad
researchpad init
researchpad runserver
# Open http://localhost:8888

Features

• Dashboard -- Real-time experiment progress, summary cards, activity feed
• Experiments -- Full history with sorting, filtering, git diffs, metric comparison
• Research -- Structured artifacts from papers, Kaggle, blogs, and GitHub
• Debug -- Deep outlier and error analysis with targeted experiment prompts
• Insights -- Theme clustering, learnings journal, diminishing returns detection
• Dark/Light Mode -- System-aware with manual toggle
• Live Updates -- WebSocket-based real-time refresh
• Keyboard Shortcuts -- Press ? for the full list

Cursor Commands

Command	Description
/experiment	Run autonomous experiment loops to improve your pipeline
/research	Research a topic and produce structured artifacts
/debug	Analyze model failures and outliers
/explain	Explain a specific experiment in detail

Requirements

• Python >= 3.10
• Node.js >= 18 (for the UI server)

Documentation

Full documentation: https://researchpad.github.io/researchpad

Development

git clone https://github.com/researchpad/researchpad.git
cd researchpad
make install
make dev

License

Apache 2.0 -- see LICENSE for details.