wanshuiyin/Auto-claude-code-research-in-sleep

ARIS ⚔️ (Auto-Research-In-Sleep) — Lightweight Markdown-only skills for autonomous ML research: cross-model review loops, idea discovery, and experiment automation. No framework, no lock-in — works with Claude Code, Codex, OpenClaw, or any LLM agent.

What it solves

ARIS (Auto-claude-code-research-in-sleep) addresses the unreliability and lack of rigor in AI-driven autonomous research. It prevents "hallucinations" and low-quality outputs by implementing a structured, audited loop of planning, drafting, adversarial review, iteration, and persistence, ensuring that AI agents can conduct deep research and code generation without constant human oversight.

How it works

ARIS is a methodology and a set of tools that can be integrated into various AI coding environments (like Claude Code, Cursor, or Trae) or used as a standalone CLI. It operates via an "audited spiral" workflow:

  1. Planning & Drafting: The agent creates a plan and an initial draft.
  2. Adversarial Review: A separate reviewer model (e.g., via Codex MCP) audits the output for integrity and fabrication.
  3. Iteration: The agent refines the work based on the reviewer's feedback.
  4. Persistence: Results are saved to a research-wiki for long-term memory and traceability.

It includes specialized tools like Anti-Autoresearch for fabrication forensics and HERO to prevent agents from "over-defending" (proposing unnecessary complexity) in their proposals.

Who it’s for

  • AI Researchers and Engineers: Those who need autonomous, high-integrity research and code generation.
  • Developers: Users of Claude Code, Cursor, Trae, or GitHub Copilot CLI who want to add a rigorous research workflow to their agents.
  • Students/Job Seekers: Specifically those preparing for AI roles (as seen in the ARIS-in-AI-Offer application).

Highlights

  • Cross-Platform Compatibility: Works as a skill-based workflow across multiple AI IDEs and CLIs.
  • Multi-Model Review: Supports cross-model auditing to ensure self-consistency and catch errors.
  • Anti-Autoresearch: A deterministic forensic tool that checks for 61 signals of AI-style fabrication and integrity hacks.
  • Multimodal Extension: ARIS-Movie-Director applies the same audited loop to generate movies from still frames.
  • Standalone CLI: Offers a full-featured CLI (ARIS-Code) with built-in support for the latest Claude models.

Related

  • Project
    Imbad0202/academic-research-skills-codex

    A Codex-native plugin that provides a rigorous framework for academic research, including specialized workflows for systematic reviews, paper drafting, and peer review simulation.

  • Project
    Masriyan/Claude-Code-CyberSecurity-Skill

    A collection of 20 Claude Code “skills” that embed cybersecurity methodologies (recon, vulnerability scanning, exploit dev, reverse engineering, threat hunting, AI‑LLM security, etc.) into Anthropic’s Claude Code assistant. Install the `skills/*` directories into `~/.claude/skills/` (or project‑local) and Claude will automatically activate the appropriate workflow when you ask a matching prompt, returning ready‑to‑run commands, reports, detection rules, or code. MIT‑licensed, Python‑based helpers, optional external tools (nmap, YARA, Ghidra, etc.), and built‑in safety gates for offensive actions.

  • Project
    CronusL-1141/AI-company

    AI Team OS is an open‑source platform that turns a Claude‑style LLM into an autonomous “CEO”. It continuously runs without prompting, orchestrating specialist agents, a task wall, research pipelines, and a two‑layer memory system. The FastAPI + React dashboard shows live workflows, decision traces, and a unified search over all memos. It’s aimed at technical founders or research teams who want an AI‑first, self‑driving company‑like workflow.

  • Project
    Dicklesworthstone/coding_agent_session_search

    A Rust‑based terminal UI and CLI that aggregates, normalises, and indexes the conversation logs of dozens of AI coding assistants into a local SQLite archive. It provides sub‑60 ms lexical search and optional on‑device MiniLM semantic search, with a stable JSON robot‑mode API for automation and diagnostics.

  • Project
    MadsLorentzen/ai-job-search

    AI Job Search is an open‑source, Claude‑Code‑driven CLI framework that automates the full job‑search funnel: it scrapes Danish (and some international) job boards, ranks postings against a structured fit rubric, drafts LaTeX CVs and cover letters, runs a reviewer LLM, verifies PDFs, checks ATS readability, tracks outcomes, and can sync to Notion or Gmail. Prerequisites include Claude Code, Python 3.10+, Bun, and a full LaTeX distribution. The repo provides detailed command definitions, portal‑scraper agents, and utilities for interview prep, skill‑gap analysis, and reporting.