Hey all, I’m a junior dev using/building Nextflow pipelines for my lab. Came from Snakemake and have been learning nf-core conventions as I go (didn’t even know

nf-core lint

existed for a while). I’ve been leaning heavily on Claude Code and found that it writes much better nf-core-compliant code when it actually has the specs in context. Without them you’re chasing lint warnings after the fact and Claude doesn’t always understand what the rule means or how to fix it properly. So I made a Claude Code skill (reusable prompt, invoke with

/nfcore-docs

) that pulls the specs and docs from nf-core/website into context automatically. Partly inspired by https://rewrites.bio/ and @Phil Ewels’ RustQC work. Still early, would love feedback: https://github.com/benjibromberg/claude-nfcore-docs

Nice! @Evangelos Karatzas has also made some skills here: https://github.com/vagkaratzas/nf-core-module-dev (he has a #C01LNCGJBMH talk about it next week: https://nf-co.re/events/2026/bytesize_modules_claude_plugin ) - would be interesting to compare notes!

TLDR for mine: nf-core-manager as a skill to orchestrate a team of 3 subagents that can work in parallel (each on their own files) to create or update an nf-core module

I wonder if it might be time to try and put together an official central set of nf-core skills.. What do you both think?

Certainly needs a discussion; I haven't made up my mind yet. On one hand it makes sense to have everything nf-core related centralised, on the other hand I feel these are my children and I can't get separated, at least not until they finish school :'D

haha, I know the feeling 😅

As a side note: especially as a junior dev I would first learn the tools before letting an llm write things for you. We have good templating, linting and docs not for llms but for humans to guide you along the way. Not learning these things will make it more difficult to be part of this community that relies quite heavily on human code reviews.

Thanks for all of the thoughtful advice! I’ll definitely check out that bytesize on

nf-core-module-dev

! @Evangelos Karatzas I’m most interested in how you actually test your skills to make sure they’re working properly since there is so much variation in llm output for any given prompt, especially when running a skill. That seemed to be the most difficult task I found when I was building my skill. That’s why I just added the option to load all the docs into context, so users don’t really have to worry as much about agents doing things correctly within the skill itself. Also, I have spent quite a bit of time learning the tools so far. The nextflow trainings (and especially the side quests) have been invaluable resources! Trying to learn as much as possible, so I’ll be at the summit in Boston next week, and I was planning on helping with a small project at the hackathon to keep learning. I mainly use LLMs as a brainstorming mechanism or a way to check my own work since I don’t work with anyone else currently who even knows nextflow, since I work with pretty much only wet lab scientists currently 😅

Best test is having the bots actually code modules, which I've been doing for some time now in secrecy...muahaha..I use hooks in my plugin to make them available to Claude and/or Codex, so they know when to use them

Cool! I’ll have to check out superpowers before your talk tomorrow!