DiffDock-L docks blind. It samples ligand placements across the whole protein surface with a diffusion model, without a search box or a declared pocket, then ranks the samples with a separately trained confidence head. Reach for it to dock a SMILES or an SDF against a PDB, or to get a starting pose for something that will rescore it. The DiffDock repository, the weights, receptor preparation and a GPU environment all have to be arranged separately; none of them are bundled here.
DiffDock confidence ranks pose plausibility, not binding affinity. A pose still needs chemistry checks and, usually, downstream scoring or refinement. Folding the receptor from sequence adds another layer of model-dependent uncertainty on top of that.
A Skill is a directory of files, so installing one is copying that directory
somewhere an agent looks. With Node 18+ and git on PATH (npm resolves a
github: spec through it), and nothing cloned:
npx github:PKU-YuanGroup/OpenAI4S install diffdock --target claude
--target claude writes to ~/.claude/skills, claude-project to
./.claude/skills, openai4s to <data_dir>/user-skills, and --dir <path>
to anywhere you name. The resolved absolute path is printed before anything is
written there, and --dry-run stops at that plan. A reinstall refuses to
overwrite a copy you have edited or one it did not install, and uninstall
removes only the files it wrote. For a recipe included in the npm release,
npx @pku-yuangroup/openai4s-skills install diffdock --target claude installs
the published copy. The npm catalog can differ from this repository.
Without Node, take the directory itself — and turn it into a .zip if an upload
field wants one. The tarball is the whole repository (over 100 MB), and the pipe
is a POSIX shell recipe (macOS, Linux, WSL) that extracts only this directory:
curl -L https://codeload.github.com/PKU-YuanGroup/OpenAI4S/tar.gz/refs/heads/main \
| tar -xz --strip-components=2 OpenAI4S-main/skills/diffdock
python3 -m zipfile -c diffdock.zip diffdock
The click-through form of that same download is the
repository zip,
which is also the route to take on Windows: extract it and copy
skills/diffdock/ out (unzip main.zip 'OpenAI4S-main/skills/diffdock/*'
unpacks only that directory). If you already run OpenAI4S there is nothing to
install — the wheel ships every bundled Skill, and a bundled Skill takes
precedence over a same-named copy in <data_dir>/user-skills. Targets,
provenance, and what the installer refuses to do:
tools/skills-installer/.
| File | Responsibility |
|---|---|
SKILL.md |
The runbook for the single-complex CLI path: how to pass SMILES, SDF and PDB inputs, what the ranked pose files and their confidence logits do and do not tell you, what hardware the run needs, and which failures are worth recognizing on sight. Geometry and affinity are kept apart throughout. |
| Directory | Responsibility |
|---|---|
references/ |
Read on demand: batch and library docking, and the sequence-only receptor path. |