Skill for structure prediction with AlphaFold 3 (AF3) from Google DeepMind. Use this skill when a user wants to predict the structure of a protein complex with ligands, DNA, or RNA; predict protein-ligand binding poses; model protein-nucleic acid interactions; use SMILES or CCD codes to specify small molecules; parse AF3 mmCIF or confidence JSON outputs; or work with AF3's structured JSON input format. AF3 handles mixed biomolecular systems that AF2 cannot model. Also trigger when the user mentions AlphaFold 3, AF3, alphafoldserver.com, protein-ligand structure prediction, PTMs (post-translational modifications), or drug-target structure modeling.
Skill for protein structure prediction and analysis with AlphaFold. Use this skill whenever a user wants to predict or fetch a protein 3D structure, download structures from the AlphaFold Database (AFDB), run ColabFold for novel proteins, parse pLDDT confidence scores or PAE (predicted aligned error) from AlphaFold outputs, predict structures of protein complexes or multimers, or work with AlphaFold3 for proteins with DNA/RNA/small molecules. Also trigger when the user mentions AlphaFold, AF2, AF3, AFDB, ColabFold, pLDDT, PAE, predicted aligned error, protein folding, or structure prediction from sequence.
Skill for biomolecular structure and binding affinity prediction with Boltz-2. Use this skill when a user wants to predict protein-ligand complex structures, estimate binding affinities (IC50/ΔG), screen compound libraries, optimize lead compounds, model protein-DNA or protein-RNA interactions, specify binding pockets as constraints, or work with cyclic peptides. Boltz-2 is MIT-licensed (commercial use allowed), supports affinity prediction alongside structure prediction, and uses YAML-based input. Also trigger when the user mentions Boltz, Boltz-1, Boltz-2, binding affinity prediction, FEP alternatives, or hit discovery screening.
Skill for biomolecular structure prediction with Chai-1 from Chai Discovery. Use this skill when a user wants to predict protein structures, protein-ligand complexes, protein-nucleic acid complexes, or multi-chain biomolecular assemblies. Also trigger when the user mentions Chai-1, chai-lab, or wants an AlphaFold3/Boltz-2 alternative that supports proteins, small molecules, RNA, and DNA. Chai-1 is Apache 2.0 licensed (commercial use allowed).
Skill for protein-ligand docking with DiffDock, a diffusion model that generates 3D binding poses from protein structure and ligand SMILES. Use this skill when a user wants to dock a small molecule to a protein, predict binding poses, run blind docking (no pocket specification required), or process batches of protein-ligand complexes. Also trigger when the user mentions DiffDock, molecular docking, binding pose prediction, or protein-ligand complex prediction.
Skill for working with ESM2 protein language models from Meta FAIR. Use this skill whenever the user wants to generate protein embeddings or representations, score variant effects or predict mutation fitness, run contact prediction, or use ESMFold for structure prediction. Also trigger when the user mentions ESM2, protein language models for embeddings, zero-shot fitness prediction, or asks to featurize protein sequences for downstream ML tasks.
Skill for generative protein design with ESM3 from EvolutionaryScale. Use this skill whenever a user wants to design or generate novel protein sequences, complete masked or partial sequences, predict 3D structure from sequence, perform inverse folding (design a sequence for a target structure), do function-conditioned generation, or iteratively refine protein designs with chain-of-thought generation. Also trigger when the user mentions ESM3, multimodal protein modeling, generative protein design, the Forge API from EvolutionaryScale, or protein sequence/structure co-generation.
Skill for efficient protein embeddings using ESM-C (ESM Cambrian) from EvolutionaryScale. Use this skill whenever a user wants to generate protein embeddings or representations, compute sequence similarities, run protein classification or regression from embeddings, cluster proteins, build a sequence similarity index, or use embeddings as features for downstream ML tasks. ESM-C is ~3× faster than ESM2 with better embedding quality — prefer it over ESM2 when embeddings are the main goal and the esm package is available. Also trigger when the user mentions ESM-C, ESM Cambrian, esmc-300m, esmc-600m, esmc-6b, or wants fast protein representations.
Skill for genomic sequence modeling and design with Evo2 from Arc Institute. Use this skill when a user wants to model or generate DNA sequences, score variant effects at single-nucleotide resolution, extract genomic embeddings, analyze mutations in non-coding or coding regions, design synthetic genomic elements, compute positional entropy across a sequence, or work with genome-scale context (up to 1M base pairs). Also trigger when the user mentions Evo2, Evo 2, Arc Institute DNA model, genomic language models, DNA variant scoring, or genome-scale foundation models. NOTE: Evo2 is a DNA model — it is NOT a protein language model. For protein embeddings or protein variant scoring use ESM2, ESM-C, or ProtT5 instead.
Skill for protein sequence-structure translation using ProstT5 from Rostlab. Use this skill when a user wants to predict protein structure as a 3Di structural alphabet string (Foldseek tokens) from an amino acid sequence, do inverse folding (recover an amino acid sequence from a 3Di structure), embed both amino acid and 3Di sequences, or work with Foldseek-compatible structure representations. ProstT5 is unique: it translates between sequence and structure via discrete tokens rather than 3D coordinates. Also trigger when the user mentions ProstT5, 3Di tokens, Foldseek sequence-structure translation, structural alphabet embeddings, or structure-conditioned protein design with discrete tokens.
Skill for inverse folding — designing amino acid sequences for a given protein backbone structure. Use this skill when a user wants to design sequences for a PDB structure, score sequences against a structure, design symmetric oligomers with tied positions, fix specific residues while redesigning others, or use evolutionary (PSSM) guidance. Also trigger when the user mentions ProteinMPNN, inverse folding, sequence design, or the Baker Lab sequence design tool.
Skill for protein embeddings using ProtTrans models (ProtT5, ProtBERT) from Rostlab. Use this skill whenever a user wants protein embeddings or representations for downstream ML tasks, protein classification (subcellular localization, membrane prediction, secondary structure annotation), similarity search, or regression tasks (stability, fitness) using Rostlab's ProtTrans family. ProtT5-XL outperforms BERT-based protein models and is competitive with ESM2 on most benchmarks. Also trigger when the user mentions ProtTrans, ProtT5, ProtBERT, prot_t5, prot_bert, Rostlab protein models, or wants HuggingFace-based protein language model embeddings that aren't ESM2.
Skill for de novo protein backbone generation with RFdiffusion from the Baker Lab (Institute for Protein Design). Use this skill when a user wants to design a new protein backbone from scratch, scaffold a functional motif into a new protein, design a protein binder against a target, generate symmetric oligomers (cyclic, dihedral, tetrahedral), redesign part of an existing structure via partial diffusion, design enzymes around an active site, create cyclic peptides, or generate diverse backbone topologies. Also trigger when the user mentions RFdiffusion, protein backbone design, motif scaffolding, binder design, protein diffusion model, or de novo protein design.
Skill for RNA sequence analysis with RNA-FM, a foundation model trained on 23 million non-coding RNA sequences. Use this skill when a user wants to embed RNA sequences, predict RNA secondary structure, classify RNA families, or analyze mRNA with mRNA-FM. Also trigger when the user mentions RNA-FM, RNA foundation model, ncRNA embeddings, RNA secondary structure prediction, or RNA language model.
Skill for single-cell biology with scGPT, a foundation model trained on 33 million human single cells. Use this skill when a user wants to annotate cell types, predict perturbation responses, integrate multi-batch scRNA-seq data, extract cell embeddings, perform reference mapping, infer gene regulatory networks, or analyze CITE-seq multi-omic data. Also trigger when the user mentions scGPT, single-cell GPT, scRNA-seq foundation models, or single-cell transcriptomics AI.
A folder with a SKILL.md file: a name, a description of when to use it, and instructions. Claude loads a skill only when the task matches its description.
How do I use one I find here?
Copy the folder into your project's .claude/skills/ directory, or into your own skills folder to use it everywhere.
What do the warnings mean?
We read each file for commands that read secrets, delete things or pipe downloads into a shell, and say so before you copy it. No warning is not a promise that a file is safe.
Which skills worked for people?
Open a skill to see its discussion. Reports from people and their agents are coming.
naitySkill