Rfdiffusion

What it does

When your agent uses it

• Designing binder scaffolds for a target protein
• Generating novel protein backbones from scratch
• Scaffolding functional motifs into new proteins
• Specifying hotspot residues for interface design

Example prompts

• “/rfdiffusion”

Requirements

• Python 3
• Docker

Workflow steps

4 steps, taken from the first numbered list in SKILL.md.

1. 1Prepare target PDB (trim to binding region + 10A buffer)
2. 2Identify 3-6 hotspot residues (exposed, conserved)
3. 3Generate 100-500 backbones
4. 4Pass to proteinmpnn for sequence design

What it can do on your machine

Read from SKILL.md and the folder at commit 59dd633. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  Shell commands in SKILL.md call:

  • python
  • conda
  • pip
  • git
  • wget

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  Hosts in commands or code, which the agent is likely to contact:

  • github.com
  • files.ipd.uw.edu

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Rfdiffusion loads about 2.3k tokens when it runs, and up to ~4k if it reads all its reference files. Until then it costs about 138 tokens; SKILL.md has 541 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~138

When it runs · the whole SKILL.md, loaded when a task matches

~2.3k

With references · SKILL.md plus every file in references/, read only if the agent opens them

~4k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from adaptyvbio/protein-design-skills at commit 59dd633, republished under its MIT licence (© adaptyvbio). 541 words, ~2,344 tokens.

name

rfdiffusion

description

Generate protein backbones using RFdiffusion, a diffusion-based generative model for de novo protein structure generation. Use this skill when: (1) Designing binder scaffolds for a target protein, (2) Generating novel protein backbones from scratch, (3) Scaffolding functional motifs into new proteins, (4) Specifying hotspot residues for interface design, (5) Creating symmetric oligomers. For sequence design after backbone generation, use proteinmpnn. For structure validation, use alphafold or chai. For QC thresholds, use protein-qc.

license

MIT

category

design-tools

tags

structure-design, diffusion, backbone, binder

proteinbase_slug

rfdiffusion

proteinbase_url

https://proteinbase.com/design-methods/rfdiffusion

git clone https://github.com/RosettaCommons/RFdiffusion.git
cd RFdiffusion

# Conda env including the required NVIDIA SE(3)-Transformer
conda env create -f env/SE3nv.yml
conda activate SE3nv
cd env/SE3Transformer && pip install . && cd ../..
pip install -e .

# Download weights (per-file hashed paths; see the repo README for the full list)
mkdir -p models
wget -P models http://files.ipd.uw.edu/pub/RFdiffusion/e29311f6f1bf1af907f9ef9f44b8328b/Complex_base_ckpt.pt

# Binder design run; single-quote the hydra args so the shell does not split [] or ,
./scripts/run_inference.py \
  inference.input_pdb=target.pdb \
  'contigmap.contigs=[A1-150/0 70-100]' \
  'ppi.hotspot_res=[A45,A67,A89]' \
  inference.num_designs=100

# De novo single chain (50-100 residues)
contigmap.contigs=[50-100]

# Binder + target (A = target chain, fixed with /0)
contigmap.contigs=[A1-150/0 70-100]

# Motif scaffolding (preserve residues, /0 = fixed)
contigmap.contigs=[20-40/0 A10-30/0 20-40]

# Multi-chain binder
contigmap.contigs=[A1-100/0 B1-100/0 60-80]

# Variable length ranges
contigmap.contigs=[A1-150/0 50-100]  # Binder 50-100 AA

# Residues for interface (chain + resnum, no spaces)
ppi.hotspot_res=[A45,A67,A89]

'contigmap.contigs=[A1-150/0 70-100]'  # Target fixed (/0), binder variable

contigmap.contigs=[A1-150 70-100]     # Missing /0 - target will move!
contigmap.contigs=[A1-150/0 70-100]   # Unquoted: shell splits on the space
contigmap.contigs=[A1-150/0, 70-100]  # Extra comma changes the contig string

'ppi.hotspot_res=[A45,A67,A89]'      # Chain letter + residue number, whole arg quoted

ppi.hotspot_res=[45,67,89]           # Missing chain letter
'ppi.hotspot_res=[A45, A67, A89]'    # Spaces inside the list break parsing

# C3 symmetric trimer
python run_inference.py \
  symmetry.symmetry=C3 \
  contigmap.contigs=[100-150] \
  inference.num_designs=50

# D2 symmetric tetramer
python run_inference.py \
  symmetry.symmetry=D2 \
  contigmap.contigs=[80-120] \
  symmetry.radius=25

# Supported symmetries: C2, C3, C4, C5, C6, D2, D3, D4, tetrahedral, octahedral

# Start from existing structure, diffuse from timestep 10
python run_inference.py \
  inference.input_pdb=initial.pdb \
  diffuser.partial_T=10 \
  contigmap.contigs=[A1-100]

output/
├── output_0.pdb       # Generated backbone
├── output_1.pdb
├── ...
└── output_99.pdb

$ python run_inference.py inference.input_pdb=target.pdb contigmap.contigs=[A1-150/0 70-100] inference.num_designs=100
[INFO] Loading model from Complex_base_ckpt.pt
[INFO] Generating design 1/100...
[INFO] Generating design 50/100...
[INFO] Generating design 100/100...
[INFO] Saved 100 designs to output/

Generated:
output/output_0.pdb (85 residues)
output/output_1.pdb (92 residues)
...

Should I use RFdiffusion?
│
├─ Need to generate protein backbone?
│  ├─ Yes → Continue below
│  └─ No, already have backbone → Use ProteinMPNN
│
├─ What type of design?
│  ├─ Binder for protein target → RFdiffusion ✓
│  ├─ De novo monomer → RFdiffusion ✓
│  ├─ Motif scaffolding → RFdiffusion ✓
│  └─ Symmetric assembly → RFdiffusion ✓
│
└─ Priority?
   ├─ Need highest success rate → Consider BindCraft
   ├─ Need diversity/exploration → RFdiffusion ✓
   └─ Need all-atom precision → Consider BoltzGen

ls output/*.pdb | wc -l  # Should match num_designs

© adaptyvbio, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 3 other files (references) in skills/rfdiffusion of adaptyvbio/protein-design-skills.

• SKILL.md
• examples/binder-walkthrough.md
• references/hotspot-selection.md
• references/parameters.md

Open the folder on GitHubat commit 59dd633

Used in 4 other repositories

We found 6 copies of this SKILL.md (exact, near-identical or edited) in other folders, from 4 other GitHub owners. This page covers the copy in adaptyvbio/protein-design-skills, which our catalogue first saw on October 7, 2026.

• majiayu000/claude-skill-registryskills/bash/rfdiffusionskills/ai-ml/rfdiffusion+1 more
• NeverSight/learn-skills.devdata/skills-md/adaptyvbio/protein-design-skills/rfdiffusion
• FreedomIntelligence/OpenClaw-Medical-Skillsskills/rfdiffusion
• BioTender-max/awesome-bio-agent-skillsskills/adaptyv/rfdiffusion