ESMFold2
SkillProductivityYour AI can predict protein structures from amino acid sequences once this skill is added. It uses fast ESMFold-style folding workflows, so predictions come back quickly without extra alignment steps. That makes it useful for triaging sequences, screening variant structures, and reviewing how confident each prediction is.
Available today. Use it from your connected AI after setup.
No other account needed.
After adding the skill, share an amino acid sequence with your AI and ask it to predict a structure, triage a set of sequences, or screen protein variants.
Then ask your AI: use the ESMFold2 skill
What your AI can do with it
- Predict protein structures from amino acid sequences
- Fold sequences quickly without building sequence alignments
- Triage sequences to decide which ones deserve a closer look
- Screen protein variants by their predicted structures
- Review the confidence behind each predicted structure
What this skill tells your AI
The instructions your AI receives, as published by companion-inc/feynman in skills/esmfold2/SKILL.md and read by ahel’s review.
Use this skill when a fast protein fold hypothesis is useful before heavier structure prediction.
Workflow:
- Normalize sequences into FASTA and record identifiers, mutations, truncations, domains, and oligomer assumptions.
- Verify the local or endpoint route before running.
- Save FASTA, model version, command or request body, predicted PDB/mmCIF, pLDDT/confidence outputs, and logs.
- Flag low-confidence regions, missing multimers, disorder, membrane regions, and sequence lengths outside the chosen route's limits.
- Use RDKit/3Dmol/PDB previews and source-backed comparisons when the output influences a research decision.
Use ESMFold-style predictions for triage unless an independent check supports the structural claim.
Signals
- GitHub stars
- 9k
- Forks
- 1k
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
esmfold2- Source
- github.com/companion-inc/feynman