whereismyai
ARIS
ARIS: Skill-based workflow and CLI for autonomous ML research, pairing an executor model with an independent reviewer model.
ARIS is an autonomous research workflow and standalone CLI that automates machine learning tasks through cross-model review loops. An executor model (such as Claude Opus or Fable) conducts the work, while an independent reviewer model (such as Codex MCP or GPT-6-Astra via a built-in bridge) audits each step. Designed for researchers and engineers, it operates across tools like Claude Code, Codex CLI, Cursor, and Trae, or as a standalone CLI. It can run without Claude or OpenAI APIs by using alternative hosts like ModelScope.
ARIS at a glance
Key features
- Execute research via a cross-model executor and reviewer architecture
- Run 83 bundled research and implementation skills
- Bridge Codex reviews automatically via codex exec without MCP registration
- Manage multi-turn sessions with resume, index lookup, and turn replays
- Support Anthropic reasoning effort levels from low to max
- Draw method figures and generate scene-checked multimodal story frames
- Inspect research fabrication signals through anti-autoresearch reporting
- Monitor session approvals via an always-on-top macOS widget
Use it for
- Automating end-to-end machine learning research and experiments
- Reviewing agent code, prompts, and implementations across distinct LLMs
- Generating clean method diagrams and multimodal image sequences
- Detecting fabrication and verifying research integrity in automated outputs
Alternatives to ARIS
- Academic Research Skills — Provides Codex-native academic research workflows and automated routing.
- OpenResearch — Run parallel research workflows locally or on remote compute.
- AI Scientist v3 — Automates end-to-end scientific research from task descriptions to complete papers.
- Math — Contains 722 mathematical manuscripts organized into 372 families.
- Arbor — It grows a hypothesis tree and tests changes on held-out data.
- FAROS — Connects literature evidence, executable plans, and human review into an auditable research loop.
- Auto Deep Researcher — Automates experiment operations to reduce manual babysitting of training loops.
- Paper2Agent — It enables direct interaction with research papers and automatic result reproduction.
- AutoResearch — It provides a stateful, reproducible workflow for automated AI research.
- Jevon Tane — It provides a rigorous, pre-registered framework for AI agents to conduct science together.
- Rayyan — Automates screening and data extraction to accelerate systematic reviews.
- Lune — Accesses trusted knowledge from top conference papers for AI agents.
All ARIS alternatives