AI Frameworks · wanshuiyin

Auto-claude-code-research-in-sleep

ARIS is a lightweight, framework-agnostic Python toolkit for automating ML research workflows through cross-model review loops, idea generation, and experiment automation. It integrates with Claude Code, Codex, and other LLM agents via Markdown-based skills, enabling autonomous research without vendor lock-in.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/wanshuiyin/Auto-claude-code-research-in-sleep

13.1k

GitHub stars

1.2k

Forks

Python

Primary language

MIT

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	wanshuiyin/Auto-claude-code-research-in-sleep
Owner	wanshuiyin
Primary language	Python
License	MIT — OSI-approved
Stars	13.1k
Forks	1.2k
Open issues	52
Latest release	v0.4.21 (2026-06-28)
Last updated	2026-07-06
Source	https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep

What Auto-claude-code-research-in-sleep is

ARIS implements research automation through composable Markdown skills, cross-model review gates, and deterministic integrity auditing. It works as a standalone CLI or integrated skill-set across multiple LLM platforms (Claude Code, Codex, Cursor, GitHub Copilot CLI), with companion tools for multimodal generation (ARIS-Movie-Director) and integrity checking (Anti-Autoresearch).

Quickstart

Get the Auto-claude-code-research-in-sleep source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep.git2cd Auto-claude-code-research-in-sleep3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Autonomous ML Research Workflows

Automate idea discovery, paper review, experiment design, and result validation using cross-model collaborative loops without manual intervention between stages.

Multi-Agent Collaborative Research

Coordinate multiple LLM agents (Claude, Codex, etc.) for peer-review style checks, integrity auditing, and self-consistency validation across research artifacts.

Research Artifact Generation & Audit

Generate and publish research outputs (blogs, cheat sheets, movie-style narratives, method diagrams) with built-in integrity checking and cross-model verification.

Implementation considerations

Requires active API keys and quota for Claude/OpenAI endpoints; costs scale with experiment volume and model complexity.
Markdown-based skill system requires learning ARIS conventions; integration time varies by target platform (Claude Code vs. Codex vs. standalone CLI).
Cross-model review loops assume multiple LLM providers available; single-provider deployments lose integrity-gate benefits.
Output quality depends on prompt engineering and agent design; results are not guaranteed reproducible across ARIS versions or model updates.
Anti-Autoresearch audit patterns (46 integrity hacks across 8 families) are research-domain specific; applicability to non-ML domains not documented.

When to avoid it — and what to weigh

Vendor Lock-In Concerns — While ARIS claims 'no lock-in', it is designed around Claude / OpenAI / Codex patterns. Portability to non-LLM workflows or proprietary agent systems may require substantial rework.
Low Tolerance for LLM Hallucination — ARIS audits output (Anti-Autoresearch catalogs integrity patterns), but fundamentally depends on LLM reliability. Critical domains (medical, safety) require human oversight and external validation.
Deterministic, Reproducible Research — LLM-based automation introduces stochasticity. If your workflow requires bit-exact reproducibility or regulatory audit trails, manual steps or classical tools may be more appropriate.
Teams Without API Access or Budget — ARIS depends on LLM API calls (Claude, OpenAI, etc.). Projects without cloud API budgets or those unable to send code/research to external services face deployment barriers.

License & commercial use

MIT License (MIT). Permissive, OSI-approved. Permits commercial use, modification, distribution, and private use without restriction. Requires attribution and inclusion of license text. No warranty provided.

MIT license permits commercial use of ARIS itself without restriction. However, usage of third-party LLM APIs (Claude, OpenAI, etc.) is subject to those providers' separate commercial terms and costs. Verify compliance with your LLM provider's acceptable-use policies and pricing model before deploying in production revenue-generating systems.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Strong
License clarity	Clear
Deployment complexity	Low
DEV.co fit	Good
Assessment confidence	High

Security considerations

ARIS itself is open-source Python; review code before use. Security posture depends on: (1) LLM API endpoint authentication and credential management (API keys should be environment variables, never hardcoded); (2) data sent to external LLM services (code, research artifacts) — confirm compliance with data residency/IP policy; (3) generated content quality — Anti-Autoresearch provides integrity auditing, not security hardening; (4) local execution sandbox (if using Claude Code or Codex) — those platforms provide execution isolation, ARIS does not. No known CVE or security incident documented in provided data.

Alternatives to consider

LangChain Agents / CrewAI

General-purpose LLM agent frameworks; broader tool integration and chaining. Less opinionated about research workflow specifics; may require more custom code for ML research automation.

AutoGen (Microsoft)

Multi-agent conversation framework with built-in role patterns. Mature ecosystem, but designed for general agent orchestration, not research-specific integrity auditing or cross-model review gates.

Continual Learning / MLflow + Human Review

Classical ML experiment tracking with human-in-the-loop validation. Deterministic, reproducible, audit-trail native. Requires manual intervention; does not automate idea generation or paper review.

Software development agency

Build on Auto-claude-code-research-in-sleep with DEV.co software developers

Evaluate ARIS for your team if you need lightweight, agent-driven research automation with integrity auditing. Start with the standalone CLI or integrate skills into your existing Claude Code / Codex environment.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

ai-berkshire generative-ai aici

Related on DEV.co

Explore the category and the services that help you build with it.

AI Frameworks AI Application Development API Development Talk to DEV.co

Auto-claude-code-research-in-sleep FAQ

Does ARIS work with my LLM provider (not Claude/OpenAI)?

ARIS is tested with Claude Code, Codex, and documented adapters for other platforms. If your provider has an agent/CLI interface and supports tool/skill calls, adaptation is feasible. Otherwise, Requires review of provider's integration surface.

What is Anti-Autoresearch, and do I need it?

Anti-Autoresearch is a separate audit tool (61 integrity signals across 8 hack families) that validates research outputs for fabrication, inconsistency, and bias. Use it if integrity is critical; ARIS workflows can run without it, but auditing recommendations omit peer-driven validation.

Can ARIS replace a human researcher?

ARIS automates workflow stages (literature review, idea generation, experiment design). Critical decisions, novelty assessment, and publication-readiness require human judgment. Designed as a productivity tool, not a replacement.

What are the cost implications?

Costs are purely API-driven: Claude, OpenAI, or other LLM provider per-token pricing. ARIS itself is free (MIT). A typical research loop (idea → experiment → audit) may cost USD 1–50+ depending on model choice, loop depth, and artifact size. Budget accordingly.

Work with a software development agency

Adopting Auto-claude-code-research-in-sleep is usually one piece of a larger software development effort. As a software development agency, DEV.co provides software development services and web development expertise — pairing senior software developers and web developers with your team to design, build, and operate ai frameworks software in production.

Automate Your Research Workflow

Talk to DEV.co Schedule a Consultation