Skip to content
Get packs

ESM protein models

Guides protein generation, ESMC embeddings and ESMFold2 folding with the Biohub esm SDK

Source Download skill
Repository
K-Dense-AI/scientific-agent-skills
Stars
55,780
Licence
MIT
Files
8 files

What it does

Teaches the agent how to use the Biohub esm Python package for protein language models: generating or completing protein sequences with ESM3, computing numerical embeddings of sequences with ESMC, and predicting 3D protein structures with ESMFold2. It explains how to choose local model weights versus hosted Biohub services, warns about mixing it with the older fair-esm package, and includes a helper script that pools per-residue embeddings into one vector per sequence.

Great for

  • Bioinformatics developers building protein sequence pipelines in Python
  • Researchers who need sequence embeddings or predicted structures for proteins
  • Teams migrating older Forge-based ESM code to the Biohub platform

Skip it if

  • Experimental validation of designs: outputs are model predictions, not lab results, and the skill says so (risk: over-trusting generated proteins)
  • Clinical or biosafety-sensitive protein design, where review by qualified experts is needed
  • Users without Python 3.12+ or the ability to install large model weights and GPU/RAM resources

Try asking

  • “Compute mean-pooled ESMC 300M embeddings for this list of protein sequences and keep the input IDs”
  • “Complete the masked positions in this sequence prompt with ESM3 and save the result as a PDB file”
  • “Predict the all-atom structure of this single-chain protein with ESMFold2 using the hosted client”
  • “Migrate my old Forge ESM3 script to the current Biohub esm SDK”
Step-by-step guide: how to use ESM protein modelsInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 8 original files, the MIT licence and an attribution note.

    esm.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip esm.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip esm.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Install esm in an isolated virtual environment separate from fair-esm, since both import as esm
  • Keep input sequence IDs and record SDK version, model checkpoint and seed with every output
  • Treat hosted API keys as secrets: read them only from the environment and never paste them into prompts

What's inside

  • references/biohub-platform.md
  • references/esm-c-api.md
  • references/esm3-api.md
  • references/forge-api.md
  • references/review.md
  • references/workflows.md
  • scripts/esm_embeddings.py
  • SKILL.md

Redistributed unmodified from K-Dense-AI/scientific-agent-skills under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.