Skip to content
Get packs

AI Research Skills Library

A collection of 98 instruction packs that teach AI coding agents how to do AI research and engineering tasks.Orchestra-Research/AI-Research-SKILLs

GitHub Download v1.7.2
Stars
14.2k
Forks
991
Licence
MIT
Language
TeX
Last update
26 days ago
Type
Curated list

What it is

This is a library of 98 'skills' organized into 23 categories, covering areas like fine-tuning, training, inference, evaluation, RAG, and paper writing. Each skill is a packaged set of guidance meant to be loaded into an AI coding agent such as Claude Code, Codex, or Gemini CLI, so the agent knows how to work with specific frameworks like vLLM, TRL, or Megatron-LM. The README says the goal is letting agents autonomously carry out research from idea to paper, with an autoresearch skill acting as the orchestrator.

  • Install the skill bundle so an agent can answer questions about frameworks such as TRL, GRPO, or vLLM with reference material
  • Ask an agent to set up a fine-tuning run using Axolotl, LLaMA-Factory, or Unsloth as described in the skills
  • Use the ML Paper Writing skills for LaTeX templates and citation checking
  • Browse the category list to find which tools cover a given stage of model development

Great for

  • Developers who already use Claude Code, Codex, or Gemini CLI and want their agent to handle ML engineering tasks
  • Researchers looking for a structured reference of ML frameworks and their common workflows
  • Students exploring which tools exist across the AI research lifecycle
  • Teams who want a shared set of agent instructions for training and serving models

Not ideal for

  • People without a coding agent installed, since the skills are meant to be loaded into one (the README does not describe use without an agent)
  • Anyone expecting a guaranteed autonomous paper-writing result; the README describes the goal but gives no evaluation of how well it works
  • Beginners who have not yet set up Python or ML environments, since the skills assume that background
Step-by-step guide: how to use AI Research Skills LibraryInstall, first run, examples, tips and FAQs

How to get it

  1. 1

    Install with a package manager

    You'll need: Node.js (required for npx; version not stated in the README excerpt) · An AI coding agent such as Claude Code, Codex, or Gemini CLI

    Interactive installer (recommended)

    npx @orchestra-research/ai-research-skills

    List installed skills

    npx @orchestra-research/ai-research-skills list

    Update installed skills

    npx @orchestra-research/ai-research-skills update
  2. 2

    Run it

    • Run the interactive installer with npx and choose the skills or bundle you want
    • Confirm the skills were installed with the list command
    • Open your coding agent and ask it a research or engineering question that matches an installed skill
  3. 3

    Get the most out of it

    • Use the Claude Code marketplace commands to install only the category you need, such as fine-tuning or inference-serving
    • Start with the autoresearch skill if you want the orchestration layer that routes to other skills
    • Keep skills current with the update command, since the library is actively maintained

Downloads

Source code (.zip)main branch · straight from GitHubRelease v1.7.2Release notes and sourceOfficial website & docsorchestra-research.com

Strengths

  • Broad coverage: 98 skills across 23 categories in one library
  • MIT licensed and open source
  • Installer supports several agents and can update or uninstall skills

Trade-offs

  • The README does not show how well the skills perform on real tasks; claims like 'production-ready' are not backed by measurements in the excerpt
  • Quality and accuracy depend on the agent and the framework versions the skills reference, which the README does not specify
  • The README is truncated, so details of many individual skills are not visible here

How it compares

ProjectStarsTypeLevelUpdatedBest for
AI-Research-SKILLs (this)14.2kCurated listintermediate26 days agoDevelopers who already use Claude Code, Codex, or Gemini CLI and want their agent to handle ML engineering tasks
Auto-claude-code-research-in-sleep18.5kAI model / agentintermediate5 days agoML researchers who already use Claude Code or Codex and want a structured research workflow
marketingskills57.9kLearning resourcebeginner6 days agoTechnical marketers who already use AI coding agents
planning-with-files27.8kDev toolbeginner6 days agoDevelopers running long multi-step tasks in Claude Code or Codex CLI
khazix-skills21.8kCurated listbeginner1 months agoDevelopers already using Claude Code, Codex, or another Agent Skills-compatible agent who want ready-made skills
OpenMythos14.9kLibraryintermediate6 months agoResearchers exploring looped or recurrent-depth transformer designs

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.