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.

Don't trust an autoresearch paper at face value. Reviewer-side integrity forensics (self-consistency + fabrication), deterministic verdict. 61 signals: 46 integrity hack-patterns (families A–H, verdict-bearing) + 13 zero-weight AI writing-style impressions (AIS) + 2 advisory. Not an opaque AI-text classifier. The dual of ARIS.

Run a Parallel Autonomous ML Research Organization on your OpenClaw instance.

Research Agora: Claude Code skills, benchmarks & tools for ML researchers — paper writing, citation verification, experiment tracking, LaTeX automation

Research-first machine learning experiment tracker for comparing model metrics, scalar curves, artifacts, and experiment lineage.

Skills that make your AI agent do ML research like a skeptical scientist, not a hype machine.

Quantum-inspired Echo State Network for multi-agent behavior analysis (MABe challenge)

Code for the paper "2Mamba2Furious: Linear in complexity, competitive in accuracy"

Autonomous ML research agent skill (Agent Skills format). Designs experiments, deploys to Modal GPUs, tracks budget, iterates autonomously.

mBFT: Metacognitive Byzantine Fault Tolerance — reference implementation for Consensus-Driven Metacognition in Multi-Agent Systems

Curated Applied Scientist & Research Engineer roles at AI labs and companies, with salaries, 2026. By Landed.

Automated Framework for Evaluating Chain-of-Thought Reasoning in LLMs.

Benchmarking PEFT (LoRA vs adapters) for ECG segment classification using Wav2Vec2, HuBERT, and ECG-FM with full metric reporting and ablations.

Provider-neutral autonomous ML research paper writer — ideation -> LaTeX writeup -> peer review -> anti-hallucination numerical audit. Spun out from hermes-sci to remove Hermes-runtime coupling.

Automate code reviews for general code projects with modular parallel checks, config-driven scope, and clear reports.

Wormhole Hyperconnections — Drop-in replacement for residual connections (x + F(x) → WHC). 50x faster at scale. 93 params. Proven stable.

Oral & Spotlight papers from top AI conferences (NeurIPS / ICLR / ICML) — curated archive for deep academic study

Hierarchical GNN for fake-news detection on NELA-GT-2022: multi-level graphs (word, sentence, article + metadata) + MLflow, baselines & explainability.

Claude Code skill for autonomous ML research — from topic to LaTeX paper

Hypothesis-first experiment tracking for agent-driven ML research, wired into Claude Code.