experiment-plan

Convert research proposals into claim-driven experiment plans with run order and success criteria.

1|1|Updated May 19, 2026
One-click install
npx skills add https://github.com/zhuyingqin/ARIS-WEB --skill experiment-plan-zhuyingqin
Or copy as Structured Prompt for Agentβ–Ό
Please help me install this Agent Skill.
Skill: experiment-plan
Source: https://github.com/zhuyingqin/ARIS-WEB/tree/main/crates/runtime/assets/skills/experiment-plan
Command: npx skills add https://github.com/zhuyingqin/ARIS-WEB --skill experiment-plan-zhuyingqin

SYSTEM DOCUMENTATION & REQUIREMENTS

πŸ’‘ This Skill includes assets (resource) components.

What problem does it solve?

It turns a refined research proposal or method idea into a claim-driven set of experiments with an execution order that a reviewer can verify.

Core Features & Use Cases

  • Claim freeze & coverage: structures primary/supporting claims and defines minimum convincing evidence and anti-claims to rule out.
  • Compact paper experiment blocks: designs a short storyline (main results, novelty isolation, simplicity/elegance, frontier necessity, and failure/diagnosis) with clear keep/cut priorities.
  • Run order + budget realism: converts the blocks into a staged milestone plan (sanity β†’ baselines β†’ main method β†’ decisive ablations β†’ polish) including cost, risk, and stop/go gates.
  • Outputs for execution: generates refine-logs/EXPERIMENT_PLAN.md and refine-logs/EXPERIMENT_TRACKER.md so execution stays organized and evidence is traceable to claims.

Quick Start

Use the experiment-plan skill after research-refine by asking it to produce a paper-ready, claim-driven experiment plan and tracker for the proposed method.

Frequently Asked Questions about experiment-plan

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I design an experiment plan for validating machine learning research claims?β–Ό

Designing an experiment plan for validating research claims requires freezing primary and supporting claims, defining minimum convincing evidence, and structuring compact experiment blocks with clear keep/cut priorities for paper defense.

What is the best way to structure ablation studies for a diffusion model paper?β–Ό

Structuring ablation studies for a diffusion model paper involves designing an ablation matrix that isolates novelty, demonstrates simplicity, and diagnoses failures, ensuring each ablation block directly maps to a specific frozen research claim.

How do I plan a compute and data budget for staged RL experiments?β–Ό

Planning a compute and data budget for staged RL experiments requires converting compact experiment blocks into a milestone plan with sanity checks, baselines, main method validation, and decisive ablations, including cost estimates and stop/go risk gates.

Can I generate a tracker for baseline selection and evaluation protocols?β–Ό

Yes, you can generate a tracker for baseline selection and evaluation protocols by outputting an EXPERIMENT_TRACKER.md file that organizes execution stages and keeps evidence traceable to your frozen paper claims.

When should I not use a claim-driven experiment roadmap for method validation?β–Ό

You should not use a claim-driven experiment roadmap when you lack a refined research proposal or method idea, as the roadmap requires extracting prior proposal artifacts to freeze paper claims and design targeted experiment blocks.

Does the experiment planning workflow support VLM research validation?β–Ό

Yes, the experiment planning workflow supports VLM research validation by applying claim-driven experiment roadmaps, ablation matrices, and evaluation protocols tailored to vision-language model method validation and paper defense.