wanshuiyin/Auto-claude-code-research-in-sleep
A structured skill workflow that runs autonomous research loops overnight in Claude Code and similar tools.
View on GitHub Updated 2026-06-16
What is wanshuiyin/Auto-claude-code-research-in-sleep?
ARIS is an open-source methodology and skill set that lets developers run structured research tasks while away from the keyboard. It implements a closed loop of plan, draft, cross-model critique, revision, and long-term memory storage so agents can tackle complex investigations without constant human oversight.
The workflow integrates directly into Claude Code, Cursor, Codex CLI and several other environments through lightweight skill files and adaptation guides. A companion agent guide makes the same process consumable by other LLMs, while extensions like ARIS-Anything broaden the loop beyond academic papers into investment, legal, and market research.
It is designed for researchers, engineers, and power users who already work inside agentic coding tools and want repeatable, auditable research output without building custom scaffolding from scratch.
What you can build with wanshuiyin/Auto-claude-code-research-in-sleep
Academic paper exploration
Feed a new arXiv paper into the loop and wake up to a reviewed summary plus follow-up questions generated through multi-agent debate.
Investment due diligence
Run the five-step process on company filings and market data to produce a persistent research wiki with integrity checks.
Long-horizon story planning
Apply the same audited workflow to visual narrative generation so each scene is critiqued before being accepted into the final sequence.
Install wanshuiyin/Auto-claude-code-research-in-sleep
Quick start
git clone https://github.com/tianyilt/claude-fleet
cd claude-fleet && bash run.sh
# open http://127.0.0.1:7878 in your browser- 1Clone the repository from GitHub.
- 2Copy the relevant skill files into your Claude Code or Cursor project.
- 3Read AGENT_GUIDE.md if you are using another LLM.
- 4Start a new session and give the agent the initial research goal.
- 5Review the generated wiki and persisted artifacts the next morning.
wanshuiyin/Auto-claude-code-research-in-sleep: pros & cons
Pros
- +Provides a clear, repeatable research loop that reduces hallucination through adversarial review.
- +Works across many popular coding agents without requiring a new platform.
- +Includes ready-made adaptations and an LLM-readable guide for quick onboarding.
- +Extends naturally to non-academic domains via the ARIS-Anything variant.
Cons
- –Still requires an existing Claude Code or compatible CLI setup.
- –Output quality depends heavily on the underlying model and prompt discipline.
- –Long-running sessions can consume significant context and API credits.
Did you find this helpful?
Frequently asked questions
No, it is a methodology and collection of skill files meant to be used inside existing agentic coding tools.
User reviews
Verified reviews from the community shape this listing's rating.
Loading reviews…
Similar agents
Other agent skills options worth comparing.
