Skip to content

Third-party Claude skill for OpenMythos — would you like a link in the README? #52

@SarthakDz

Description

@SarthakDz

Hi Kye,
I'm a final-year engineering student in India. I read through OpenMythos over the last couple of days and wrote a Claude skill that gives Claude careful knowledge of the repo — the eight architectural invariants (ρ(A) < 1, e frozen across loops, MoE only in the recurrent block, etc.), the debugging playbook, the variant-scaling rules, and the papers you cite. When it's loaded, Claude behaves like a senior engineer who has actually read main.py, instead of giving generic PyTorch advice.

Repo: https://github.com/SarthakDz/OpenMythos-Skill

A few things I want to be upfront about:
It's a third-party harness, not affiliated. The README credits you prominently and links back here.
All substance comes from your work and the papers you cite. The skill is just the packaging.
I'd love to add a link in your README if you think it's useful. Equally happy to drop the project entirely if you'd rather this not exist.
Thanks for OpenMythos — the writeup on LTI stability and loop-index embedding is one of the cleaner explainers I've read.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions