ARIS

A lightweight, Markdown-only set of skills for autonomous machine learning research, adding cross-model review loops, idea discovery and experiment automation to any coding agent without a framework or lock-in.

Agent skills

Visit website github.com

ARIS (Auto-Research-In-Sleep) is a research methodology packaged as Markdown-only skills rather than a platform, so it can be used inside Claude Code, Codex CLI, Cursor, GitHub Copilot CLI, OpenClaw or any other LLM agent. One model, the executor, drives the research work while an independent reviewer model, Codex by default, checks it, so review and generation are not done by the same model.

Its skills cover cross-model review loops, idea discovery, and experiment automation for machine learning research. It is also distributed as a standalone CLI and as agent plugins, in addition to being installed directly as skills.

It is written in Python and released under the MIT license.

More in Agent skills

See all 581

Provenance gate
github.com
A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.
- Agent skills
- AI evals
Expected Harm Calculus (EHC-4): Asimov's Four Laws for Autonomous AI Agents
github.com
A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.
- Agent skills
Vitruvius
github.com
An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.
- Agent skills
- AI evals

More in Agent skills

Provenance gate

Expected Harm Calculus (EHC-4): Asimov's Four Laws for Autonomous AI Agents

Vitruvius

Cookies