ESMFold (Claude Skill)

Predicts protein structure directly from a single sequence with ESMFold — no MSA required — driven as a Claude skill.

   
Type Claude Skill
Supplier Meta AI / EvolutionaryScale
Availability GA — Claude Science research skill
Pricing Free / OSS (MIT code; model weights per Meta AI terms)
Capabilities Read/Write — writes predicted structures (PDB) with pLDDT
Verified works · 2026-07-20
Security caution · 2026-07-20 — featured Claude Science skill, MIT, but self-host facebookresearch/esm repo is archived/unmaintained

How to install

  • Claude Science — enable the built-in ESMFold research skill (Anthropic-hosted; not published to the public anthropics/life-sciences marketplace).
  • Run the model yourself — the upstream model is open source:
    git clone https://github.com/facebookresearch/esm
    

    Follow the repo README for environment setup and model weights.

What it does

Runs ESMFold (built on the ESM-2 protein language model) to fold single sequences without an MSA. Uses local inference, or the ESM Atlas fold API (api.esmatlas.com/foldSequence/v1/pdb/) for quick one-off predictions.

Primary use cases: Fast single-sequence folding, high-throughput proteome folding, orphan-protein modelling

Notes

Claude Science: Featured as a research skill in Anthropic’s Claude Science. Its inclusion there is an independent signal of quality and trustworthiness for life-science work.

For embeddings and generative ESM3 / ESM C design see the broader ESM skill; this entry covers ESMFold structure prediction.

Sources


Installed this tool?

Share feedback — install path, OS, errors, workarounds. The form opens with this tool pre-selected and a link back to this page.