Experiment Plan

wanshuiyin/auto-claude-code-research-in-sleep

This turns a research method into a runnable experiment roadmap with a clear claim-to-evidence structure. It's designed for the point where you've settled on an approach and need to figure out what experiments actually defend the paper, in what order, with what baselines. The workflow locks down your primary claims first, then builds a compact set of experiment blocks that test novelty, simplicity, and whether any LLM or frontier component is genuinely necessary rather than decorative. You get a full execution plan with milestones, compute estimates, and a separation between must-run and nice-to-have experiments. The skill defaults to three seeds and caps you at five core blocks to keep the story tight. Use it after research-refine or when you need an ablation matrix and run order that a reviewer would actually respect.

skills.sh installs: 578
GitHub stars: 18.9K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add wanshuiyin/auto-claude-code-research-in-sleep --skill experiment-plan --agent claude-code

Installs into .claude/skills of the current project.

Files

Experiment Plan: Claim-Driven, Paper-Oriented Validation

Refine and concretize: $ARGUMENTS

Overview

Use this skill after the method is stable enough that the next question becomes: what exact experiments should we run, in what order, to defend the paper? If the user wants the full chain in one request, prefer /research-refine-pipeline.

The goal is not to generate a giant benchmark wishlist. The goal is to turn a proposal into a claim -> evidence -> run order roadmap that supports four things: