Built on AI-Research-SKILLs · Open Source · MIT

The AI research skill layer

98 open skills across 23 categories — from model architecture to paper writing — that install into Claude Code, Codex, Gemini CLI, or Cursor with one command.

98
Skills
23
Categories
4
Agents Supported
1
Command Install
The basics

What’s a skill?

A skill is a focused SKILL.md — expert patterns, quick recipes, and reference docs your agent loads only when it needs them. Install one, and your agent knows a framework cold: the right commands, the common gotchas, working config — no pasting documentation into the chat.

SKILL.md
A tight playbook: when to use it, the core patterns, the pitfalls.
Reference docs
Deeper material your agent pulls in progressively, only as needed.
One command
Install into your agent; it auto-loads when the task calls for it.
Skill Catalog
Browse all skills
Filter by category or search by name
00Autoresearch

autoresearch

Orchestrates end-to-end autonomous AI research projects using a two-loop architecture. The inner loop runs rapid experiment iterations with clear optimization targets. The outer loop synthesizes results, identifies patterns, and steers research direction. Routes to domain-specific skills for execution, supports continuous agent operation via Claude Code /loop and heartbeat monitoring, and produces research presentations and papers. Use when starting a research project, running autonomous experiments, or managing a multi-hypothesis research effort.

View skill.md
01Model Architecture

implementing-llms-litgpt

Implements and trains LLMs using Lightning AI's LitGPT with 20+ pretrained architectures (Llama, Gemma, Phi, Qwen, Mistral). Use when need clean model implementations, educational understanding of architectures, or production fine-tuning with LoRA/QLoRA. Single-file implementations, no abstraction layers.

View skill.md
01Model Architecture

mamba-architecture

State-space model with O(n) complexity vs Transformers' O(n²). 5× faster inference, million-token sequences, no KV cache. Selective SSM with hardware-aware design. Mamba-1 (d_state=16) and Mamba-2 (d_state=128, multi-head). Models 130M-2.8B on HuggingFace.

View skill.md
01Model Architecture

nanogpt

Educational GPT implementation in ~300 lines. Reproduces GPT-2 (124M) on OpenWebText. Clean, hackable code for learning transformers. By Andrej Karpathy. Perfect for understanding GPT architecture from scratch. Train on Shakespeare (CPU) or OpenWebText (multi-GPU).

View skill.md
01Model Architecture

rwkv-architecture

RNN+Transformer hybrid with O(n) inference. Linear time, infinite context, no KV cache. Train like GPT (parallel), infer like RNN (sequential). Linux Foundation AI project. Production at Windows, Office, NeMo. RWKV-7 (March 2025). Models up to 14B parameters.

View skill.md
01Model Architecture

distributed-llm-pretraining-torchtitan

Provides PyTorch-native distributed LLM pretraining using torchtitan with 4D parallelism (FSDP2, TP, PP, CP). Use when pretraining Llama 3.1, DeepSeek V3, or custom models at scale from 8 to 512+ GPUs with Float8, torch.compile, and distributed checkpointing.

View skill.md
How It Works
Three steps. No setup.
From zero to expert AI research agent in minutes.
01

Choose a category

Browse 23 research categories — fine-tuning, mech interp, distributed training, paper writing, and more.

02

Pick your skills

Each skill is a focused SKILL.md with expert patterns, quick recipes, and progressive reference docs.

03

Install — pick your path

Path A: CLI for developers (Claude Code, Cursor, Gemini CLI). Path B: Upload SKILL.md to Claude chat. Both fully supported.

$npx ai-research-skills install
Read the full guide
Why Aires
Built for serious research.
Purpose-designed for production AI research workflows.

98 Skills

Full AI research stack — architecture, training, inference, interpretability, RAG, multimodal, paper writing.

Zero setup

One command installs any skill into Claude Code, Codex, or Gemini CLI. No configuration files needed.

MIT License

Fully open source. Use commercially, fork it, extend it. No restrictions, no lock-in.

Agent-native

Follows Anthropic's official skill format — tight SKILL.md with progressive-disclosure reference docs.

Aires mascot
Get started now

Start with Autoresearch

The orchestrator skill that coordinates all others — from literature survey to paper submission.

npx ai-research-skills install autoresearch