AI-Research-SKILLs — Installieren & Live-Demo

Warum nutzen

Hauptfunktionen

87 skills across 22 categories: model architectures, tokenization, fine-tuning, post-training, distributed training, optimization, inference, safety
Autoresearch orchestrator skill manages full experiment lifecycle
References folder per skill with official docs, GitHub issues, tutorials
Covers modern post-training (TRL, GRPO, SimPO, verl, slime) and inference (vLLM, SGLang, TensorRT-LLM, llama.cpp)

Live-Demo

In der Praxis

ai-research-skill.replay ▶ bereit

0/0

Installation

Repository des Autors · Offizielle Dokumentation zu Claude Skills

Anwendungsfälle

Praxisnahe Nutzung: AI-Research-SKILLs

Fine-tune a Llama model with Unsloth + LoRA without Googling

👤 ML engineers running fine-tunes on a single GPU box ⏱ ~20 min intermediate

Wann einsetzen: You have a dataset and want Claude to produce a runnable training script with correct flags.

Voraussetzungen

Unsloth installed — pip install unsloth — the skill will remind you of the nightly build caveats

Ablauf

Describe the target run

Use the unsloth skill. Fine-tune Llama-3-8B on my dataset at ~/data/train.jsonl, QLoRA, 3 epochs, output to ./out.✓ Kopiert

→ Claude writes a script with correct Unsloth imports and real CLI flags
Ask for the config explanation

Walk me through each hyperparameter and why.✓ Kopiert

→ Reasoning grounded in Unsloth's docs, not generic ML clichés

Ergebnis: A training script that runs on the first try.

Fallstricke

Claude mixes HF Trainer and Unsloth idioms — Insist on 'use the unsloth skill only' — don't pull from Transformers skill

Kombinieren mit: filesystem

Set up a vLLM server with correct tensor-parallelism for your GPUs

👤 Infra engineers serving LLMs in production ⏱ ~30 min advanced

Wann einsetzen: You have 2-8 GPUs and want Claude to pick the right --tensor-parallel-size and --max-model-len.

Ablauf

State the hardware and model

Use the vllm skill. Serve Qwen2.5-72B on 4x H100s. Give me the exact launch command and sanity tests.✓ Kopiert

→ Correct TP size and quantization recommendation
Ask for load-test script

Now give me a locust or vllm-benchmark script to verify throughput.✓ Kopiert

→ Runnable benchmark using the right endpoint format

Ergebnis: A vLLM deployment with sanity checks and a benchmark baseline.

Fallstricke

Claude picks a TP size that doesn't divide the attention heads — The vLLM skill references list valid TP sizes per model family — have Claude cite that

Kombinieren mit: aws

Draft an ML paper with correct structure and citation conventions

👤 Researchers writing for NeurIPS/ICML/ICLR ⏱ ~60 min advanced

Wann einsetzen: You have experiment results and need a paper outline that matches the venue's style.

Ablauf

Provide results and venue

Use the ml-paper-writing skill. Target ICLR. Here are my results [paste]. Draft the intro, method, and experiments sections.✓ Kopiert

→ Structure follows venue conventions, ablation tables properly framed
Revise for reviewer concerns

What would Reviewer 2 push back on? Add preemptive responses.✓ Kopiert

→ Concrete weaknesses, not generic 'add more experiments'

Ergebnis: A draft that passes initial sanity review and is worth polishing.

Fallstricke

Claude over-claims in the abstract — Explicitly tell it to mirror the restraint of top-tier accepted papers

Kombinieren mit: arxiv

Kombinationen

Mit anderen MCPs für 10-fache Wirkung

ai-research-skill + filesystem

Store training configs, logs, and checkpoints while Claude drives experiments

Save the run config under experiments/<date>-<name>/ and tail the logs.✓ Kopiert

ai-research-skill + arxiv

Pull related work while the ml-paper-writing skill drafts sections

Find 5 recent arXiv papers on GRPO and have the paper-writing skill weave them into the related work section.✓ Kopiert

Werkzeuge

Was dieses MCP bereitstellt

Werkzeug	Eingaben	Wann aufrufen
Autoresearch orchestrator	research goal	Complex multi-phase experiments
Fine-tuning (Axolotl, LLaMA-Factory, Unsloth, PEFT)	dataset + base model	SFT / LoRA / QLoRA runs
Post-training (TRL, GRPO, OpenRLHF, SimPO, verl, slime)	reward model or preference data	Alignment, preference optimization
Distributed training (DeepSpeed, FSDP, Megatron, Accelerate)	model + cluster topology	Multi-node or multi-GPU training
Inference (vLLM, SGLang, TensorRT-LLM, llama.cpp)	model + hardware	Serving a model efficiently
Optimization (Flash Attention, bitsandbytes, GPTQ, AWQ, GGUF)	model weights	Fitting a model on smaller hardware

Kosten & Limits

Was der Betrieb kostet

API-Kontingent: None for the skill itself
Tokens pro Aufruf: Heavy — SKILL.md + references can load 5-10k tokens per sub-skill
Kosten in €: Free — skills are local files; you pay for compute you run
Tipp: Scope prompts to one sub-skill at a time; the full 87-skill library is too big to load simultaneously.

Sicherheit

Rechte, Secrets, Reichweite

Credential-Speicherung: No credentials — skill is prompts + references

Datenabfluss: None from the skill itself

Some referenced frameworks have rapid API churn — confirm against the current version if a call fails.

Fehlerbehebung

Häufige Fehler und Lösungen

Claude blends two frameworks (e.g. Accelerate + DeepSpeed) incorrectly

Name one skill in the prompt and tell Claude to ignore the other.

References feel outdated

The repo is TeX-heavy and may lag the bleeding edge. For brand-new frameworks, supplement with a fresh fetch of the official docs.

Skill not auto-invoking on a relevant prompt

Mention the framework by name — 87 skills overlap and auto-routing is fuzzy.

Alternativen

AI-Research-SKILLs vs. andere

Alternative	Wann stattdessen	Kompromiss
scientific-agent-skill	You need biology/chemistry/clinical, not ML training infra	Different domain focus
huggingface MCP	You want live HF Hub operations rather than expert prompts	MCP gives you real API actions; this skill teaches the patterns

Mehr

Ressourcen

📖 Offizielle README auf GitHub lesen

🐙 Offene Issues ansehen

🔍 Alle 400+ MCP-Server und Skills durchsuchen