What the tech

Search

Browse topics

/ opens search from anywhere. Esc closes it.

Tools
What the tech
Get the newsletter

ARIS (Auto Research In Sleep): a Claude Code skill by Yang Ruofeng

For ML researchers who want agents to plan experiments, review drafts with a second model and outline papers. Skip it if you don't want drafts sent to another provider.

Before you install: The paper-plan skill allows unrestricted Bash and sends your outline to an outside reviewer model by default.

Copy install command View the source
Score
73of 100
Rank
#143of 239
Installs
41.9Kon skills.sh
Stars
19Kon GitHub
Security
Clear 4 scanners
Updated
2026
Context
~54tokens always

Install

npx skills add https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep

Run it in your terminal. View the source

Most people install it straight from GitHub, where it has 19K stars, so its skills.sh count understates how widely it is used.

Using the Claude app rather than Claude Code? You can add a skill there too. How to add it in the app

What does ARIS (Auto Research In Sleep) do?

ARIS (Auto Research In Sleep) is a set of 87 Markdown skills for running machine learning research with little supervision. Members cover idea discovery, novelty checks, experiment planning and queues, result analysis, auto review loops, citation audits, rebuttals, slides and posters; some trigger phrases are in Chinese. The paper-plan member reads NARRATIVE_REPORT.md, AUTO_REVIEW.md, experiment results and idea reports, then builds a claims and evidence matrix, a section plan sized to a venue such as ICLR or NeurIPS, a figure plan and citation scaffolding. It sends the outline to a second model for review, by default an OpenAI model through Codex MCP.

It comes as a set of skills that work together:

  • ablation-planner (534 installs)
  • alphaxiv (486 installs)
  • analyze-results (580 installs)
  • arxiv (598 installs)
  • auto-paper-improvement-loop (569 installs)
  • auto-review-loop (597 installs)
  • auto-review-loop-llm (522 installs)
  • auto-review-loop-minimax (516 installs)
  • citation-audit (522 installs)
  • citation-crosscheck (46 installs)
  • claims-drafting (512 installs)
  • comm-lit-review (374 installs)

When should you use ARIS (Auto Research In Sleep)?

Use it when

  • You have results and need a venue-ready paper outline
  • You want another model to critique your plan
  • You run long experiment loops overnight

Skip it when

  • Your drafts must not leave your machine
  • You don't work on ML papers

What it needs and what it can touch

You need

  • A Codex MCP server or another configured reviewer model
  • Python 3 for helper tools

What it can do

  • Runs shell commands, including Python helper scripts
  • Reads and writes project files
  • Sends outlines to an outside model (OpenAI through Codex MCP by default)
  • Searches and fetches web pages

From reading its instructions, not from running it.

Why it scores 73

Strongest on upkeep (100 of 100), weakest on adoption (51).

  • Adoption35 points 51
  • Upkeep20 points 100
  • Trust20 points 65
  • Quality15 points 100
  • Reader answers10 points Not enough answers yet, so this does not count

How the formula works

Is ARIS (Auto Research In Sleep) safe to install?

Every scanner that checked it passed it. Scanners often miss what other scanners catch, and a skill can change after a scan. Read its source before you install it.

ScannerResultDate
Gen Agent Trust HubPassedNov 10, 2026
SocketPassedNov 10, 2026
SnykLow riskNov 10, 2026
ZeroLeaksPassedApr 16, 2026

Details

Installs
41,889
Repository stars
18,971
Last repository change
Tokens always loaded
about 54
Tokens when used
about 5,294
License
MIT
Data confidence
High

Alternatives to ARIS (Auto Research In Sleep)

All skills for this job